Electoral Violence in Sub-Saharan Africa: Why Peace Processes Fail 9781626375406

Citation preview

ELECTORAL VIOLENCE IN SUB-SAHARAN AFRICA

ELECTORAL VIOLENCE IN

SUB-SAHARAN AFRICA Causes and Consequences

Stephanie M. Burchard

Published in the United States of America in 2015 by FirstForumPress A division of Lynne Rienner Publishers, Inc. 1800 30th Street, Boulder, Colorado 80301 www.rienner.com

and in the United Kingdom by FirstForumPress A division of Lynne Rienner Publishers, Inc. 3 Henrietta Street, Covent Garden, London WC2E 8LU

© 2015 by Lynne Rienner Publishers, Inc. All rights reserved

Library of Congress Cataloging-in-Publication Data Burchard, Stephanie M. Electoral violence in sub-Saharan Africa: Causes and consequences / by Stephanie M. Burchard. Includes bibliographical references and index. ISBN 978-1-62637-251-1 (hardcover : alk. paper) 1. Elections—Africa, Sub-Saharan. 2. Political violence—Africa, Sub-Saharan. 3. Democracy—Africa, Sub-Saharan. 4. Africa, Sub-Saharan— Politics and government—21st century. I. Title. JQ1879.A5B87 2015 303.60967—dc23 2015026425

British Cataloguing in Publication Data A Cataloguing in Publication record for this book is available from the British Library.

This book was produced from digital files prepared by the author using the FirstForumComposer.

Printed and bound in the United States of America

The paper used in this publication meets the requirements of the American National Standard for Permanence of Paper for Printed Library Materials Z39.48-1992.

5 4 3 2 1

Contents

List of Tables and Figures Acknowledgments

vii ix

1

The Violence of Voting in Africa

2

The Political Context of Electoral Violence

23

3

Kenya: Strategic Electoral Violence

55

4

Senegal: Incidental Electoral Violence

85

5

Liberia: Postconflict Electoral Violence

103

6

Electoral Violence and Voter Turnout

123

7

Electoral Violence and Democratic Attitudes

147

8

Electoral Violence and the Future of Democracy in Africa

167

Appendix: Fear of Electoral Violence and Willingness to Vote Bibliography Index

v

1

175 177 189

Tables and Figures

Tables

1.1: Types of Electoral Violence

13

2.1: Pre-Electoral Violence and Rule of Law & Accountability

44

2.2: Pre-Electoral Violence and Executive Elections, Ordered Logit Regression

46

2.3: Pre-Electoral Violence and Legislative Elections, Ordered Logit Regressions

47

3.1: Electoral Violence and Vote Choice in Kenya’s Provinces, 2007–2008

65

3.2: Patterns of Competition and Pre-Electoral Violence in Kenya’s Districts, 2007

67

3.3: Patterns of Competition and Pre-Electoral Violence in Kenya’s Constituencies, 2007

68

3.4: Ethnicity and Fear of Political Violence in Kenya, 2008

70

3.5: PID and Ethnicity in Kenya, 2002–2008

72

6.1: Voter Turnout and Pre-Electoral Violence in Africa, Voting Age Population

129

6.2: Voter Turnout and Pre-Electoral Violence in Africa, Isolated Sample

130

6.3: Fear of Electoral Violence in Africa

134

6.4: Willingness to Vote in the Future

135

6.5: Fear of Electoral Violence and Willingness to Vote

137

6.6: Fear of Electoral Violence and Willingness to Vote in Africa, Random Effects Logit Regressions

139

vii

viii

Electoral Violence in Sub-Saharan Africa

6.7: Impact of Fear of Electoral Violence on Willingness to Vote in the Future in Kenya, Senegal, and Liberia

141

7.1: Fear of Electoral Violence and Democratic Disposition, Random Effects Ordered Logit Regressions

157

7.2: Electoral Violence and Democratic Disposition in Africa, 2008–2009 (Predicted Probabilities)

158

7.3: Electoral Violence and Democratic Disposition in Kenya, 2008–2009 (Predicted Probabilities)

159

7.4: Political Trust and Fear of Electoral Violence in Africa

160

7.5: Political Trust and Fear of Electoral Violence in Kenya

161

Figures

1.1: A Decade of Electoral Violence in Africa, 2004–2013

3

2.1: Incidences of Pre-Electoral Violence in Sub-Saharan Africa, 1990–2014

37

2.2: Pre-Electoral Violence and Electoral Rules, Executive Elections

38

2.3: Pre-Electoral Violence and Electoral Rules, Legislative Elections

39

2.4: Pre-Electoral Violence and Type/Timing of Elections

41

2.5: Pre-Electoral Violence and Democracy (Binary)

42

2.6: Pre-Electoral Violence and Democratic Quality

43

2.7: Nonviolent Elections in Africa, 1990–2014

49

6.1: Voter Turnout and Pre-Electoral Violence in Africa, Registered Voters

129

6.2: Voter Turnout and Pre-Electoral Violence in Kenya, 2007

132

7.1: Fear of Electoral Violence in Africa, 2008–2009

153

7.2: Fear of Electoral Violence in Africa, 2011–2013

154

Acknowledgments

This book would not have been possible without the assistance of many, many people. First and foremost, I am grateful for the tireless advice and suggestions offered by Dorina A. Bekoe. Your mentorship and friendship have been invaluable. It’s been an honor to learn from you. I have also received immeasurable assistance from colleagues Janette Yarwood, Ashley Bybee, Caroline Ziemke-Dickens, Katy Hassig, Nancy Jones, Ben Lindorf, Tom Milani, Herman Philips, Rick Wright, Rick Porterfield, Phil Major, and George Ward. A special thanks to David S. Chu, who has been a champion of this work since the beginning. Very capable research assistance came from Ivana Djak, Andrea Pongo, Betty Boswell, and Brittany Gregerson. Ashton Callahan, I thank you for your data collection assistance, travel companionship, and very detailed notes. A hearty thank you to Scott Straus and Charlie Taylor for their initial data collection efforts, which inspired this project. Thanks for comments and feedback at various conferences to Jacki Klopp, Norma Kriger, Manuela Travaglianti, Leonardo Arriola, Gabriel Bardell, and James Long. And, a big thank you to colleagues Jacki Wilson, Debi LiangFenton, and Linda Bashir at the United States Institute of Peace. I was fortunate to receive assistance from several experts in country while conducting field research. In Kenya, I was assisted by Felix Odiambo, the staff at EISA, and William Otienga. In Liberia, Ezekiel Pajibo, Senesee Freeman, and Oscar Bloh gave generously of their time and expertise. In Senegal, Ousmane Sene and Doudou Dia were great sources of wisdom and insight. The staff at Lynne Rienner Publishers went above and beyond in providing assistance and feedback. I owe a huge debt of gratitude to Lynne Rienner and Alejandra Wilcox in particular. Thanks also to the two anonymous reviewers, who assisted me in vastly improving the initial manuscript. Finally, on a personal note, I would like to extend a grateful thank you to my family and friends who have provided words of support and encouragement over the past three years. To Toni Burchard, Pamela Mason, Adley Grace Mason, Damon Mason, Colin ix

x

Electoral Violence in Sub-Saharan Africa

Anderson, Kristen Schauer Klas, Jacqueline Chamberlain, Michelle Romo, Taras Matla, David Ashton, Michael Scott, Philip Owira, Jen Blum, Andrew Spiegelman, Amelia Scholtz, Evan Robertson, Chris Bean, and Tiffany Barnes: thank you, thank you, thank you. To my colleagues and instructors at the University of Cape Town, Bob Mattes, Shaheen Mozaffar, the late Joel Barkan, Erica Penfold, Moletsane Monyake, and Carlos Shenga, I am forever grateful. To anyone I overlooked, I’m sorry, I’m forgetful, and I thank you.

1 The Violence of Voting in Africa

On December 30, 2007, it was announced that Mwai Kibaki had been re-elected president of Kenya, the victor of a close and hard-fought election. Jubilation quickly turned to fear as mere minutes later protesters took to the streets in cities around the country, some armed with rocks and pangas. Supporters of Kibaki’s chief rival, Raila Odinga, cried foul, claiming that the election was rigged. Within a few hours, violence had spread like wildfire as security forces clashed with scores of protesters in the capital Nairobi, as well as smaller towns such as Kisumu, Eldoret, Mombasa, and Molo. Within days, the country was in chaos. Fighting continued for two months, despite repeated domestic and international pleas for a return to peace and calm. The violence finally ended in February 2008 once Kibaki and Odinga signed a power-sharing agreement, but not before more than 1,300 Kenyans had lost their lives, hundreds of thousands had lost their homes, and countless were injured. The postelection violence, or PEV as it has come to be known, shook Kenya to its core and placed it at risk of succumbing to the same fate as so many of its neighbors—civil war. Compared to many other African countries, Kenya was largely considered a “success” story up until the events of 2008, its post-colonial history marked by solid economic growth and political stability under single party rule. While neighboring countries such as Sudan, Uganda, Ethiopia, and Somalia experienced wars, rebellions, and coups d’état, Kenya experienced continuity of rule through the successive tenures of founding president Jomo Kenyatta and his hand-picked successor, Daniel arap Moi. In 1991, the government repealed its ban on opposition parties, a development that many interpreted as a move toward democracy. After winning the 1992 and 1997 presidential contests, Moi stepped aside in accordance with the constitution’s provision on executive term limits. The 2002 election resulted in an opposition victory and the defeat of Moi’s party, the long-ruling Kenya Africa National Union 1  

2

The Violence of Voting in Africa

(KANU). A coalition of smaller parties with Kibaki as its leader decisively defeated KANU’s candidate, Uhuru Kenyatta, with 61.3 percent of the vote to Kenyatta’s 30.2 percent. The Carter Center’s 2002 delegation noted this “milestone for democracy” and congratulated the country on its “democratic and peaceful transfer of power.”1 It is against this backdrop that Kenya’s PEV took the world by surprise. Kenya's success was weighted by its political stability. By adding in other parts of the story, such as adherence to rule of law, respect for human rights, or level of corruption, Kenya's story loses its luster. And if one were to examine the quality of its electoral contests, Kenya would certainly not have been considered such a success. Almost all of Kenya’s multiparty elections have been violent, with death totals at times in the thousands. Even the country’s most recent elections held in 2013, heralded as a triumph of democracy and a declared a resounding success by many observers including the International Foundation for Electoral Systems (IFES), were violent.2 At least 300 Kenyans died prior to the election as a result of politicking and campaigning.3 The election violence that wracked the country in 2007 and 2008 was, unfortunately, not an anomaly—not for Kenya and not for subSaharan Africa. As I detail in this book, more than half of all elections held in Africa experience some form of violence or intimidation either before or after election day. And Kenya, although on the extreme end of the spectrum, is not the only country in Africa where intense violence routinely takes place during elections. Large-scale violence frequently accompanies elections in the Democratic Republic of Congo, Nigeria, and Zimbabwe as well. Multiparty elections are a relatively recent phenomenon in Africa. During the post-colonial period, most African countries were ruled by various forms of dictatorship and autocracy. Coups and instability were common. Only Botswana, Mauritius, and The Gambia experienced extended periods of peaceful multiparty elections. Beginning in the early 1990s, a seismic shift in regime type took place. In quick succession, the majority of countries in sub-Saharan Africa moved away from authoritarianism and toward multiparty electoral regimes. Benin was one of the first to transition to multiparty elections in 1990. By 1997, almost 75 percent of the sub-continent had adopted multiparty elections.4 The spread of electoral regimes across Africa was welcomed and largely celebrated by the international community as a sign that the subcontinent was moving toward democracy. However, the nearly universal embrace of elections across the African sub-continent—only Eritrea has not held multiparty elections since 1990—has brought with it an unanticipated and troubling trend: violence. Since 1990, more than 50 per-

Electoral Violence in Sub-Saharan Africa 3

cent of African elections can be characterized as violent with voters experiencing harassment, intimidation, and in some cases, death, as a direct result of the electoral process. A quick look at the past decade of elections held in Africa demonstrates the nagging persistence of election violence. Figure 1.1 depicts the percentage of elections held between 2004 and 2013 in sub-Saharan Africa where some form of violence took place. From 2004 to 2008, there was substantial year-to-year variation in the frequency of election violence occurring, with some years much more violent than others. Since then, the trend line seems to be decreasing but still remains close to 50 percent.

Percent

Figure 1.1: A Decade of Electoral Violence in Africa, 2004 to 2013

90 80 70 60 50 40 30 20 10 0 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Percent of Violent Elections

Electoral violence has, in certain instances, catalyzed prolonged political conflict. Take for instance Côte d’Ivoire’s 5-month dispute over the 2010 presidential election in which an estimated 3,000 were killed and hundreds of thousands displaced.5 Angola also represents a worst case scenario of elections igniting conflict: multiparty elections held in 1992 as part of a conflict resolution process and as a means of introducing democracy led to the resumption of a civil war which did not end until a decade later. Although relatively rare, the fact that electoral violence has been associated with the outbreak or resumption of larger conflicts suggests we urgently need to better understand why it occurs, what it affects, and what measures can be taken to prevent it.

4

The Violence of Voting in Africa

The 2007-2008 PEV in Kenya was the culmination of years of problematic elections and lingering tensions between politicians, parties, and ethnic groups. The 2002 elections brought into power a coalition of opposition parties but within just two years the coalition had fallen apart amid accusations that the president had not honored an informal powersharing deal with his partners. In November 2005, a constitutional referendum that was endorsed by President Kibaki was defeated, with 58 percent voting against the proposed constitution to 42 percent in favor. Raila Odinga of the Liberal Democratic Party (LDP), who prior to the referendum had been a coalition partner and minister in the Kibaki government, campaigned extensively against the government’s version of the constitution. Shortly after the referendum’s defeat Kibaki dismissed his entire cabinet. When he announced his new cabinet two weeks later, Raila and the LDP were conspicuously absent. Raila and several other opposition leaders used the momentum of the “No” campaign to create a new coalition, the Orange Democratic Movement (ODM), in order to contest the next elections. By 2007, the animosity between Kibaki and Raila had hardened to an unmanageable degree, but it served as only the most recent polarizing divide in a country where the individual has often trumped party, policy, and ideology. Ethnicity in particular has loomed large during Kenya’s political history. The Kikuyu are the largest ethnic group in Kenya, comprising between 17 and 22 percent of the population. Despite representing at most approximately one-fifth of the population of Kenya, there have long been accusations that the Kikuyu have been privileged over other the country’s other ethnic groups. The Luhya and Luo compete for position as the second and third largest ethnic groups. The Kalenjin are estimated to be the fourth largest ethnic group in Kenya, comprising approximately 12 percent of the total population. Kenya’s first president Jomo Kenyatta, who served from 1963-1978, was Kikuyu. Many claimed that he played ethnic favoritism in both politics and business dealings. In Rift Valley specifically, it has been alleged that the Kikuyu disproportionally benefited from Kenyatta’s government. Many Kikuyu were sold arable farmlands at favorable prices and at the expense of other groups who had historical claims to the lands.6 Kenyatta’s first vice president, Oginga Odinga, was a prominent Luo chieftain. After approximately two years in office, Oginga had a falling out with Kenyatta and resigned. He then established the Kenya People’s Union (KPU) in opposition. In 1969 the government banned the KPU and Oginga was imprisoned for two years. He remained a vocal member of the opposition until his death in 1994.

Electoral Violence in Sub-Saharan Africa 5

Upon Kenyatta’s death in 1978, his vice president Daniel arap Moi, appointed after Oginga’s defection, assumed office and served as president until 2002. Moi was Kalenjin. During Moi’s presidency it was alleged that he showed ethnic favoritism, appointing Kalenjin to various important ministerial posts and administrative positions but also promoting Kikuyu over other groups so as to not rock the boat.7 The lead up to the 2007 election was tense and hate speech predicated on ethnic stereotypes was rampant. Vernacular radio stations were particularly problematic as some used their broadcasts to call for violence against or the displacement of specific ethnic groups.8 One of the informal campaign slogans of the 2007 election was “41 against 1,” an indirect reference pitting the Kikuyu (led by Kibaki) against the countries remaining 41 ethnic groups. Member of Parliament (MP) William Ruto, a long-standing KANU member from Eldoret North and a Kalenjin, defected from the party after his bid to run for the 2007 presidency was not supported by former president Moi and other KANU leaders. He joined ODM and endorsed their presidential candidate Raila Odinga, son of former vice president and long-time opposition leader Oginga Odinga. In the months before the elections, political parties were intimidating voters and orchestrating forced displacements around the country. For example, residents of Molo, a town in Rift Valley Province were explicitly told to vote Raila or leave.9 Leaflets were dropped around the region stating: Onyo! Onyo! Onyo! Warning! Warning! Warning!...A warning has been issued to the people who are not from this region! This is our land from before!...Time has come for you to leave our land and return to yours!...Whoever disobeys will die! The Rift Valley Land Owners and Protectors army is ready to fight for its right till the last blood drop is shed!10

Molo, comprised of Kikuyu, Kisii, and Kalenjin peoples, had a population of around 100,000 in 2007. In previous elections, Molo had voted in favor of the opposition, but the margins had been diminishing over time. In 1992, the opposition parliamentary candidate won with 73 percent of the vote; in 2002, the opposition candidate won the Molo constituency with 54 percent of the vote. Molo also had a history of pre-election violence. Prior to the 1992 and 1997 elections, there were significant Kalenjin and Kikuyu antagonisms believed to have been orchestrated by former president Moi and his supporters.11 Prior to the 2007 election, at least 16 people were killed

6

The Violence of Voting in Africa

and 300 families displaced in Molo.12 According to a state-sponsored report conducted in the aftermath of the PEV, youths were ferried in from neighboring towns for the explicit purpose of intimidating voters. The report also found that local politicians were responsible for distributing the leaflets.13 The National Security Intelligence Service (NSIS) allegedly had information as early as October 2007 that Kalenjin ODM leaders were planning to attack Kikuyu residents in Kuresoi, Olenguruone, and Molo due to their suspected support of President Kibaki.14 The NSIS also had information that Zakayo Cheruiyot, MP for Kuresoi, had hired Kalenjin youth to intimidate and displace Kisii and Kikuyu prior to the election.15 In assessing blame for the violence the Waki Commission, which was impaneled in February 2008 to investigate the causes and perpetrators of the PEV, specifically admonished the provincial administration for ignoring warning signs that violence in Molo was imminent.16 After months of tension and violent flare-ups, the official vote took place on December 27, 2007. The tallying process, however, took longer than anticipated which led some to speculate that the election was being tampered with. Three days later, on December 30, 2007 at 5:50pm local time, the Electoral Commission of Kenya (ECK) announced behind closed doors with only the state-run media present that Kibaki was reelected president. He was sworn in almost immediately. Mere moments after the results were announced violence broke out across the country. 17 In Rift Valley, Kalenjin raiders, backed by some prominent politicians in the area, began attacking and burning down Kikuyu homes.18 Kalenjin youth were brought into the area, housed and fed by Kalenjin politicians and businessmen in Molo.19 Kikuyu and Kisii retaliated by responding in kind. In turn, more Kalenjin youth/warriors were brought in from Kericho and Ndoinet.20 In a speech televised live the next day, Odinga refused to concede defeat and accused the government of rigging the elections in Kibaki’s favor. The government responded by shutting down Odinga’s press conference and indefinitely suspending all live television and radio broadcasts.21 That same day ECK commissioners, using Molo as an example of potential voting irregularities, requested that an independent inquiry be established to investigate electoral fraud. On 2 January, Samuel Kivuitu, chair of the ECK, told reporters that he had made the December 30th pronouncement of the victor under duress. Kivuitu told the press that he wasn’t sure who had actually won the election.22 For more than a month after the election, riots and protests continued around the country. Rift Valley was especially hard hit. On January 24th, approximately 400 Kikuyu youths from Molo armed mostly with

Electoral Violence in Sub-Saharan Africa 7

pangas decided to attack Sirikwa Farm in Kuresoi in the heart of Kalenjin country.23 Kalenjin leaders heard of this plan and waged a counter attack, killing 13 Kikuyu before they were able to attack Sirikwa.24 In total, 150 people were killed in Molo district, 170 people were injured, 1,564 houses were burned down, and 66,000 people were displaced—all as a result of violence that took place before and directly after the election.25 It took weeks of negotiations and international mediation for the violence to subside. Ultimately, a government of national unity was established that included Kibaki and Odinga as president and prime minister, respectively. The agreement also called for the writing of a new constitution and a re-assessment of the electoral process in Kenya. Implicit in these decisions was an acknowledgement that the existing institutions and electoral mismanagement both contributed to the post-election violence. According to the EU Election Observation Mission in its official assessment released April 2008, there was at least a 20,000-vote discrepancy between the constituency tally and national tally in favor of Kibaki in Molo alone.26 Senegal

Similarly to Kenya, post-colonial Senegal has often been considered one of the sub-continent’s more successful countries. Despite a low-level insurgency in the southern region of Casamance that has taken place since the early 1980s, the country has enjoyed stability and peace since independence. After an orderly transition to independence from French colonial rule in 1960, the Socialist Party (PS) remained in power until 2000. Léopold Sédar Senghor, founder of the PS, ruled the country until his voluntary resignation on January 1, 1981. Senghor’s prime minister, Abdou Diouf, then took over as president with little controversy. Diouf won presidential elections in 1983, 1988, and 1993. In 2000, long-time opposition candidate Abdoulaye Wade won the second round of presidential elections, garnering 58.5 percent of the vote to Diouf’s 41.5 percent. Unlike several other African leaders, Diouf readily accepted the outcome and left office. This election was regarded by many as an indicator of Senegal’s democratic progress. Due in no small part to the country’s history of peaceful transfers of political power, many considered Senegal to be an exemplar case of post-colonial state management and democratic development.27 Thus Wade’s attempted power grab ahead of the 2012 elections took many by surprise. When faced with expiring term limits—which he himself had helped to implement—Wade “re-interpreted” the constitution so as to

8

The Violence of Voting in Africa

stay in power. He also endeavored to modify existing electoral rules to ensure an easy victory. Wade’s extra-constitutional maneuvering beginning in 2011 led to protests around the country. Many in the opposition were concerned that Wade would not allow himself to lose the 2012 election and would stay in power by any means necessary. These concerns were magnified in January 2012 by increasingly hostile interactions between Wade’s security forces and protesters. Wade’s security forces and opposition protesters clashed on several occasions, causing injuries and a handful of deaths. As the February presidential election approached, tensions continued to mount. Protests devolved into riots. In a country with a reputation for harmonious resolution of its political disputes, the pre-election clashes that took place in 2012 were startling. Even more alarming, as many as 15 fatalities were reported as a result of interactions between protesters and government security forces. Some began to fear that Senegal was on the verge of significant unrest. There seemed to be no way to peacefully resolve the crisis. The situation appeared intractable. Wade was signaling that he would go to extreme ends to win the election and the opposition remained steadfast that they would not allow this to happen. And then, almost overnight, the violence subsided. Protesters returned to their homes, security forces returned to their barracks and opposition members began preparing for the second round of elections. To what can we attribute the rapid escalation of pre-election tensions and their quick denouement? I argue that protesters were directly responding to the signals that President Wade was sending to the populace. Once Wade announced that a second round of elections was necessary—a significant concession on his part—the opposition knew that Wade could be defeated. This knowledge led to a de-escalation of tensions and a return to calm. Political elites often set the tone for the conduct of elections. What they say, and what they do not say, has a huge impact on the eruption of violence. As I will demonstrate, the Senegalese case underscores how the transparent management and observation of elections can reduce tensions over the electoral process and promote a peaceful transfer of power. The Senegalese case is also instructive in that the management of elections has undergone significant reform over time, but this progress has been the result of push and pull interactions between the government in power and various opposition groups. The process has been piecemeal and at each phase, the ruling government has sought to undermine the intended effects of reform.

Electoral Violence in Sub-Saharan Africa 9

Liberia

Unlike Kenya and Senegal, Liberia’s recent history includes significant conflict and civil war. Beginning with a military coup in 1980, Liberia experienced a period of prolonged social and political unrest that did not end until 2003. Unlike most other countries in Africa, Liberia was never under colonial rule. It was established in the mid-1800s by freed American slaves and was governed from 1847 until 1980 by a small elite political class descended from these early settlers. The True Whig Party won every election held in Liberia from 1878 until a master sergeant in the army deposed the government of then-president William Tolbert. Samuel K. Doe ruled in an increasingly repressive fashion until he was killed in 1990 by insurgent forces leading a rebellion against his government. War between different rebel factions raged on until a ceasefire was reached in 1996. Elections were held in 1997 as a part of the conflict resolution process. Charles Taylor, leader of one of the larger rebel factions, was elected president. While elections were deemed “free and fair” by international observers, most agree that the vote was less a true competition over political power and more a vote against the resumption of war. Taylor was a ruthless dictator and in 1999 the Second Liberian Civil War began by armed groups opposed to his rule. In 2003, as international pressure against Taylor’s regime increased, he finally relented and fled Liberia for exile in Nigeria. A transitional government was installed to prepare the country for a return to multiparty elections. The United Nations (UN) deployed a peacekeeping force with some 15,000 troops. In 2005, Liberia held its first postconflict elections. Ellen Johnson Sirleaf, who had run against Taylor in 1997, won the second round of presidential elections, although her opponent George Weah disputed the outcome. She was re-elected in 2011 amid an opposition boycott of the second round of balloting. Elections in 2005 and 2011 were close affairs. Fraud was alleged in both cases but election violence was kept to a minimum. The non-occurrence of electoral violence in such a fragile environment, when many other indicators would have suggested that it should have taken place, requires further explanation. I argue that part of the reason lies with the heavy presence of the international community. The 2005 elections were almost entirely managed by the international community, which provided significant logistical and technical assistance. UN peacekeeping forces provided election security alongside the Liberian National Police. Although the government of Liberia took up much of the responsibility for organizing and administering the 2011 elections, UN peacekeepers remained in country. The international community

10

The Violence of Voting in Africa

has, in effect, taken up many of the key functions of election management for the past two elections. Much like elections in Senegal, capable and impartial management can go a long way to preventing election violence from erupting. However, just below the surface lie myriad grievances and substantial fear of violence on the part of the Liberian electorate. There is little guarantee that future elections will continue to be peaceful. Case Selection

In this book, I describe in detail the electoral experiences of Kenya, Senegal, and Liberia. When analyzed together, these three countries provide a comprehensive picture of the nature of electoral violence in Africa. Kenya is a country in which electoral violence often takes place. The violence, although varying in intensity over time, is frequently organized at the highest levels of government and with the specific purpose of affecting election outcomes. Violence has occurred at multiple points in the electoral process, both before and after Election Day. In Senegal, moderate violence has broken out sporadically and under very specific circumstances. It is a case in which election violence is less intentional but potentially just as damaging. The 2012 election in particular appeared to have the ability to throw the country into significant political disarray. Liberian elections, on the other hand, have routinely been violence-free. Although many other indicators might suggest that Liberia’s postconflict elections should have resulted in violence, peace has reigned. These three cases compliment and help to concretize the quantitative analysis presented in Chapters 2, 6, and 7 of this book. They also exhibit temporal variation and country-specific variation, allowing for some interesting comparisons. These three cases were selected so that the full array of types, timing and outcomes of election violence could be examined. They were also selected because one of the key factors often associated with electoral violence—the electoral system—is relatively similar across all three cases. However, because I am interested in exploring the causes and, just as importantly, the consequences of electoral violence these three cases also illustrate different experiences with electoral violence which will allow for a more nuanced examination of how electoral violence (or its absence) affects democratic development, the subject of the second half of this book.

Electoral Violence in Sub-Saharan Africa 11

Types of Electoral Violence

Electoral violence, while not an Africa-specific phenomenon, is more pronounced on the continent. Globally, it is estimated that violence occurs in roughly 19 percent of elections, with riots and protests accompanying approximately 14 percent of elections.28 In contrast, violence and intimidation occur in approximately 58percent of elections in Africa. Even more alarming is the fact that since the proliferation of multiparty elections in the early 1990s the vast majority of African countries—86 percent—have experienced electoral violence. Fortunately, extreme cases such as Kenya’s post-election violence are relatively rare—affecting no more than 10 percent of elections held in Africa—but unfortunately, this is not a trend that has substantially decreased over time. The persistence of electoral violence, even in less deadly forms, may have a significant effect on the development of democracy on a continent where democracy as we know it is a relatively new phenomenon. Complicating things even further, democracy advocates often encourage the adoption of multiparty elections as a means of reducing conflict and promoting political stability. Democracy is often believed to present a peaceful alternative to autocratic rule. Whereas in autocracies there are no mechanisms for the population to influence government behavior short of rebellion or an overthrow of the government, democracy allows individuals periodic opportunities to communicate their preferences to their leaders through regularly scheduled elections. As such, many argue that democratic regimes should be more stable and less violent than autocratic regimes. In the early 2000s, however, researchers such as Snyder (2000) and Fischer (2002) began to take note of the propensity for new democracies and postconflict countries to experience significant violence either as part of a nation-building exercise or as a consequence of fraudulent and untrusted processes.29 In 2009, Paul Collier concluded that pseudodemocratic countries, ones in which elections are held but without any of the other substantive features of democracy such as a free press and adherence to the rule of law, often experience more generalized political violence than some of their autocratic counterparts.30 Electoral violence is a subset of political violence but it has several distinct features that differentiate it from other forms of political violence. It differs in terms of its timing and intent—namely, influencing electoral outcomes. Coups, rebellions, and repression all take place with no regard to electoral processes. In a 2009 article, Höglund makes this particular distinction, arguing that differences in motive, timing, actors, activities, and targets allow us to separate electoral violence from these

12

The Violence of Voting in Africa

other forms of political violence.31 This point is not trivial as the motivation and means of electoral violence are unique to its ends. Electoral violence intends to affect the outcome of an election; political violence intends to affect a variety of political outcomes ranging from specific policy decisions to outright regime change. Additionally, the timing and motive of electoral violence is tied to scheduled elections and as such is inherently more predictable but also potentially more persistent since elections are by nature recurring. Electoral violence encompasses any intimidating or harassing action that is directly related to the electoral process. It may take place prior to an election, on Election Day, or immediately after an election has taken place, often as a result of the announcement of the outcome. The definition of electoral violence includes a range of behaviors that includes: the distribution of hate-speech leaflets, the forced displacement of specific groups of voters, political assassinations, and targeted violent attacks. It also encompasses protests and riots that occur as a direct result of elections. Although seemingly disparate acts, these behaviors have one thing in common—they are meant to affect the outcome of an election through force. Electoral violence is distinct from other forms of politicized violence in that actors use the existing electoral framework in order to achieve their goals. For example, a coup d’état, a common form of political violence, is focused at the elite level and involves the forced removal of a party or individual from political power. Actors generally have no regard for existing laws and as such it is an unpredictable and extrajudicial act. Electoral violence, however, works within existing and established timelines to achieve the acquisition of political power. Electoral violence also shifts the focus from elite level actors to include voters as potential targets of violence. Although sometimes unplanned (as has been the case in countries such as Guinea and Senegal), electoral violence is often times mobilized by political actors (as in Kenya, Zimbabwe, and Nigeria). Because of this strategic purpose of electoral violence, political institutions can either constrain or encourage its employ. Complicating things, however, due to the illicit nature of electoral violence political actors have a vested interest in obscuring their role, thus at the time what may seems spontaneous is actually deliberately planned and managed. Electoral violence can be subdivided into its different forms based on intent, method, timing, target, and actors. Table 1.1 outlines a basic typology of electoral violence. Assuming the likely motivation of the actors, we can separate electoral violence into two distinct categories: incidental and strategic. Incidental electoral violence occurs as a product

Electoral Violence in Sub-Saharan Africa 13

of protest around electoral events, either before or after an election. It takes place when tensions are heightened and groups from opposing sides are in close proximity to each other. It may be triggered by the perception (real or imagined) of electoral impropriety. Incidental electoral violence is not strategic. It is not pre-planned but rather a spontaneous occurrence. Typically, this type of violence involves protesters and perhaps over-zealous security forces—as was the case in Senegal in 2012—but it may also arise out of mutual frustration between supporters of rival candidates or parties. Table 1.1 Types of Electoral Violence Type

Methods

Timing

Perpetrators

Incidental

Protests, Riots

Before, Day of, After

Opposition members, Protesters, Security forces

Strategic:

Threats, Physical attacks, Assassinations, Bombings, Forced displacement

Before, Day of

Politicians and their agents

IEDs, Bombings, Riots

Before, Day of, After

Actors excluded or marginalized from the electoral process

Suppressive or Mobilizing

Strategic: Disruptive

Strategic electoral violence, on the other hand, is pre-planned. It is deliberately employed to affect the outcome of an election. Strategic electoral violence can be used for a multiplicity of electoral purposes including suppressing voter intention, mobilizing voter support, or disrupting an election altogether. The first two types of strategic electoral violence, suppressive and mobilizing, can be conceived as part of a larger electoral strategy a candidate or party employs in an attempt to deliberately affect an outcome. Although some argue that electoral violence is primarily meant to suppress voter turnout, I present evidence that shows that violence has been used to suppress turnout as well as to mo-

14

The Violence of Voting in Africa

bilize voters. Both activities can achieve the same goal: the distortion of citizen preference so as to reduce the competitiveness of an election and win office. Politicians in Kenya have used violence to both suppress and mobilize votes, sometimes in the same election. In 1992, it is believed that Kenyan president Moi and his party supporters killed 1,500 and forced the displacement of approximately 250,000 residents in Rift Valley in order to prevent them from voting for the opposition.32 Moi won the 1992 with 36.4 percent of the total vote and a margin of victory of approximately 500,000 votes. In the lead up to the 2007 elections some residents of Rift Valley stated that they were told to vote for Raila Odinga or else, with the “else” being interpreted as a veiled threat. In the Kenyan context, it is an easy message to interpret. This type of violence is mobilizing, using threats to force voters to cast their ballot for a certain candidate. Political assassinations, those that target both aspirants and activists alike, can either be suppressive or mobilizing depending on the context. Assassinations in the most direct sense of the act suppress candidacies but they may also serve to mobilize voter support through fear of future attacks. Disruptive violence is meant to prevent a vote from taking place or to change an already-announced outcome. The perpetrators may be marginalized electoral actors seeking to expand their influence beyond the agreed upon electoral arena. If the allegations in the Waki report are to be believed, members of Kenya’s political elite organized the electoral violence after the 2007 elections in order to force the ruling party to enter into a governing coalition with the announced losing party, ODM. Actors outside of the electoral process, such as rebel groups or terrorist organizations, also perpetrate disruptive electoral violence. In Kenya’s Coast Province, the Mombasa Republican Council (MRC), a small separatist movement seeking independence, waged a pre-election campaign in 2013 meant to prevent the election from ever taking place. Their stated goal was to deprive the process of legitimacy in order to demonstrate lack of support for the existing governance structures.33 Similarly, in Nigeria the terrorist organization Boko Haram was accused of engaging in electoral violence as a means of discouraging voters from participating in the 2011 and 2015 elections. Because these actors are generally external to the process, acts perpetrated by groups such as Boko Haram and the MRC are more akin to terrorist attacks. As such, their activities are largely outside of the scope of this book. In theory, strategic pre-election violence is meant to influence voter behavior (namely vote choice and/or voter turnout) before an election. Displacement, intimidation, and targeted political assassinations are the

Electoral Violence in Sub-Saharan Africa 15

most frequent tactics used to influence an electoral outcome through suppression or mobilization; however, if pre-election violence were to become too pervasive, too obvious, and perhaps too deadly, it could derail an election altogether as the government or even external actors would be expected to intervene. If the goal is electoral disruption, this is not a problem but if the goal is to strategically influence voter behavior and hence, an outcome in favor of one candidate or party over another, violence entrepreneurs must find a equilibrium that achieves influence but does not require immediate attention or intervention. In contrast, post-election violence occurs after an election and is used as a way to either punish victors and their supporters or, if strategically employed, used to force victors into negotiations with losers to share political power. Because this method is extra-judicial and outside of the purview of the accepted rules of the electoral “game”, levels of violence must reach such a level that it forces action on the part of the winning party. As such, election violence after the fact should be significantly more intense than violence that takes place before an election. The Kenyan case conforms to these expectations. Although these categories are conceptually and theoretically distinct, as was stated before, strategic electoral violence can at the outset appear to be incidental as its entrepreneurs seek to affect the outcome of an election with as little recrimination as possible. The inherent desire to evade punishment, even if it is a remote possibility due to weak criminal justice systems, often makes the exact measurement and quantification of electoral violence difficult. Thus it is necessary to combine quantitative and qualitative approaches in the study of electoral violence. Empirical analysis allows us to broadly identify patterns and relationships associated with the causes and consequences of electoral violence that hold across countries and time periods, but it does not always allow for a detailed analysis of the different types or motivations behind electoral violence. Contextualizing electoral violence in specific electoral environments allows us to better identify the actors and likely culprits behind its use and to assess what these actors might hope to gain from it use. The case studies in this book do just that. For example, in the immediacy of the post-election violence in Kenya in 2008, it was believed that the violence was a spontaneous reaction to a close and contentious election. The investigations that took place after the violence subsided, however, revealed a more complicated picture, one where opportunists from both main presidential candidates’ camps organized violence. The losing side planned violence in order to leverage itself into a post-election governing coalition. The government then responded in kind., this tactic was successful in forcing a govern-

16

The Violence of Voting in Africa

ment of national unity be formed to quell the violence. Thus the postelection violence can be considered disruptive as it resulted in the use of extra-judicial means to force a change in the outcome of the elections and how elections are conducted in the future. All of these types of violence can take place within a single election, and in the case of Kenya, they have. Electoral fraud is intimately related to electoral violence in several ways but the nature of the relationship is determined by the type and timing of violence. Fraud and election violence often occur in the same election. Fraud affects post-election violence as post-election violence is a response to an electoral outcome. Perceived unfairness, manipulation or the belief that fraud is likely to take place can trigger an emotive response prior to an election which can translate into incidental violence. However, electoral fraud is not necessarily a trigger for strategic violence—the type of violence that requires organization and planning— but rather it is a companion strategy. The same politicians who are willing to win at any cost—including using violence and intimidation to influence votes—also frequently employ fraud to win. Fraud and violence are both illicit strategies that complement each other as they are both meant to achieve the same goal—ensuring electoral victory. Elections in Senegal have rarely experienced strategic violence, but fraud, manipulation, and incidental violence have frequently occurred. The role of election fraud is common to the experiences of all three countries examined in this book. Although it was a confluence of factors that led to the 2007 and 2008 post-electoral violence in Kenya, electoral fraud was a contributing cause. The opposition insisted that widespread irregularities and vote tampering took place that amounted to the theft of the presidency by the incumbent Kibaki and his supporters. In both the 2012 Senegalese elections and the 2013 elections in Kenya, however, the widespread presence of election observers made fraud and violence untenable electoral strategies. Unlike Kenya, however, the electoral reforms that have been adopted in Senegal were adopted incrementally, giving actors and institutions time to get used to their new roles. While fraud has been alleged in Liberia, there has been insufficient evidence to substantiate these claims. In order to better contextualize electoral violence, Chapter 2 describes in greater detail the patterns of electoral violence in Africa through a systematic examination of the political and institutional factors associated with its occurrence. As will be detailed, executive and legislative elections both experience significant levels of electoral violence, suggesting that violence is not just the purview of strongman presidents as some have asserted. Additionally, electoral institutions and

Electoral Violence in Sub-Saharan Africa 17

the political environment under which elections are held affect both the occurrence of pre-election violence and which variant (strategic or incidental) is most likely to take place. Chapter 3 delves further into the nature of electoral violence in Kenya, beginning the 1960s and continuing through the most recent elections held in 2013—the first general election to be held after the PEV. Because the Kenyan experience encompasses so many of the different types of electoral violence in terms of timing, intent, and severity, it makes an excellent case study. Additionally, because strategic violence has occurred quite frequently in Kenya, a closer look at Kenyan elections allows me to uncover the factors that encourage or impede the use of strategic violence in particular. Chapters 4 and 5 put electoral violence in a comparative perspective by extending the analysis to Senegal and Liberia in order to examine how and why incidental forms of electoral violence occur and what factors may prevent new democracies on the precipice from descending into widespread violence. The 2012 Senegalese elections, although nowhere near as violent as those in Kenya, were arguably the most violent in that country’s history with multiple fatalities, hundreds of injuries, and countless protests. Many were concerned that mounting violence could lead Senegal down a dangerous path; however, the electoral violence eventually subsided and calm returned to the country. The reasons why have to do largely with the quality and conduct of elections. Over the course of multiple decades the opposition agitated for reforms that improved the transparency of elections and reduced opportunities for fraud, culminating in the peaceful resolution of the 2012 election crisis. The Senegalese case may be instructive for other newly democratizing countries seeking to improve electoral management. Liberia’s experience with elections is complicated and the outcomes in terms of peace are varied. The country has held elections since the mid-1800s but competition was restricted to an elite ruling class who rarely shared power. After civil war broke out in 1989, elections were organized in 1997 as a means to resolve the conflict. A fragile peace was constructed but voting was far from free or fair. There was an implicit threat of the resumption of war if one of the candidates did not win the election, transforming it from a simple contest between politicians into a referendum on peace. The peace, unfortunately, lasted less than two years before civil war began again. Elections were held in 2005 after another peace accord was struck but these elections were managed much differently and the outcome was much more positive. Peace has endured since and another relatively calm electoral contest took place in 2011. The past two Liberian elections appear much more peaceful than those

18

The Violence of Voting in Africa

in Senegal and Kenya but lying just beneath the surface are grievances which could threaten to derail the progress that has thus far been achieved. Conclusions

Although from a normative perspective, we may believe that electoral violence is undesirable it is still not known exactly how electoral violence affects the democratization process. Electoral violence has the potential to cause serious instability and conflict. It may even lead to adverse regime change. In its most extreme form, it has triggered the resumption of war, as in the case of Angola. Even in its less intense forms, electoral violence may bring about the weakening of representational ties between politicians and voters and may ultimately undermine the consolidation of democracy. By examining effects on individuals, upon whose consent democracy depends, I am better positioned to assess this relationship. Chapters 6 and 7 analyze how electoral violence affects voter turnout and attitudes toward democracy. The analytic lens is broadened and the experiences from several other African countries are included in cross-national analyses to explore the effects of electoral violence on voter behavior and democratic dispensation. Electoral violence seems to have a null effect in the aggregate on voter turnout. This is likely due to the multitude and sometimes crossed purposes of electoral violence and the difficulty in assessing when exactly coercive voting is taking place (i.e., casting a ballot for a candidate or party that one would not support in the absence of violence). Electoral violence is, however, significantly related to stated willingness to vote in future elections, but this relationship is complicated and based on partisan attachments and the type of electoral violence that takes place. Electoral violence is also significantly related to individual assessments of democracy, support for democracy, and trust in governing institutions. Chapter 8 builds upon the previous chapters to address why electoral violence is so problematic and what can be done to prevent it. Some have argued that electoral violence serves as a catalyst to needed changes in the conduct of elections and leads to improvements in democratic quality over time.34 To them, while violence is not necessarily a desirable part of the electoral process, it is a normal part of the democratization process and to be expected as countries seek to get the institutions right. There are two main problems with that approach this book seeks to address. First, electoral violence can be traced to very specific environments, institutions, and triggers that are associated with its

Electoral Violence in Sub-Saharan Africa 19

prevalence. Because we can identify antecedent causes we can also identify specific and targeted responses to prevent it from taking place. When electoral violence takes place, it is the intervention after the fact and NOT the violence that is important. Second, electoral violence in new democracies may be common but it is not preordained, nor is it a necessary part of the democratization process. It may be a byproduct of transitioning away from a more authoritarian regime to a more open and liberal regime but this does not mean that it cannot be prevented. Additionally, while there are instances of seemingly spontaneous or emotive violence that have manifested around elections, there are also many instances in which political actors mobilize supporters or hire gangs to terrorize electoral actors. If actors are deliberately organizing violence, then efforts to prevent them from doing so can be pursued. Even in cases in which the violence is decentralized and there is no deliberate organization for its perpetration, it is possible to identify ways in which tempers could have been reduced and violence could have been mitigated. Furthermore, over the past few decades electoral violence in Africa appears to be more frequent and more intense than other types of pollitically-motivated violence that may occur between elections. Since 1997, election years in Kenya, Senegal, and Liberia have been significantly more violent than years without elections. According to the African Conflict Location and Event Dataset (ACLED), between 1997 and 2014, Kenya experienced on average 210 violent incidents in non-election years.35 During election years, the average number of violent incidents increased to 360. The average number of fatalities per year also increased—by almost double—from an estimated 330 in non-election years to 640 fatalities in election-years. Even controlling for the 2007 election, which could be argued to be anomalous due to its intensity, the average number of fatalities in Kenya during election years was 430. In Senegal and Liberia, a similar pattern repeats as election years experienced, on average, significantly more violent incidents than nonelection years. Although the violence was lower in frequency and intensity in both countries compared to Kenya (closer to 50 incidents in each country per year and with fatalities relatively rare), there is still a significant difference between election years and non-election years in terms of violence. In Senegal, election years experienced an average of 54 violent incidents compared to 42 in non-election years.36 In Liberia, the number of incidents almost tripled during election years, from 23 in non-election years to 61.37 Elections may be displacing other patterns of political violence and channeling incidents and efforts into specific peri-

20

The Violence of Voting in Africa

ods of time so as to take advantage of the electoral cycle. This is a very troubling trend. Some have argued that flawed elections, including those in which violence takes place, when held in unbroken succession will eventually result in advances in freedoms and civil liberties and can contribute to the democratization process. I argue that flawed elections undermine the process necessary for democracy to take hold and to flourish.38 When voters come to associate elections with predictable patterns of violence, their appetite and support for democracy eventually wanes. The acceptance of electoral violence as a normal part of the democratization process suggests that there is nothing that can be done prior to electoral violence taking place but these interventions and remedies could have absolutely been adopted prior to the violence. Violence is a blunt, imprecise tool. In some cases it can create new grievances and can spiral out of control. Additionally, because it masquerades as a part of the democratic process, it can have serious deleterious effects on democratic development in terms of attitudes and behaviors of the people it is meant to protect. Ignoring or accepting electoral violence may, in the end, undermine democracy to the point where the international community, rather than promoting democratic development, is actually preventing it from taking root. Notes

                                                                                                            1

“Observing the 2002 Kenya Elections.” Sweeney, Jr., “Kenya’s 2013 Elections: an Effective Assistance Model?” 3 “High Stakes: Political Violence and the 2013 Elections in Kenya.” 4 Adejumobi, “Elections in Africa.” 5 “They Killed Them Like It Was Nothing.” 6 Boone, “Land Conflict and Distributive Politics in Kenya.” 7 Lynch, I Say to You. 8 “Kenya: Spreading the Word of Hate.” 9 Burchard, “The Resillient Voter?” 10 “Kenya: Clashes, Elections and Land.” 11 Lynch, I Say to You. 12 “The Waki Report.” 13 “The Waki Report.” 14 “The Waki Report.” 15 “The Waki Report.” 16 “The Waki Report,” 85. 17 Gettleman, “Disputed Vote Plunges Kenya into Bloodshed.” 18 “The Waki Report.” 19 “The Waki Report.” 20 “The Waki Report.” 2

Electoral Violence in Sub-Saharan Africa 21

 

21

“Odinga Rejects Kenya Poll Result.” Berger, “Kenya’s Poll Chief Does Not Know if Kibaki Won.” 23 “The Waki Report.” 24 “The Waki Report.” 25 “The Waki Report,” 89. 26 “Kenya: Final Report General Elections December 27, 2007,” 34. 27 Galvan, “Political Turnover and Social Change in Senegal.” 28 Norris, “Why Electoral Malpractices Heighten Risks of Electoral Violence.” 29 Snyder, From Voting to Violence: Democratization and Nationalist Conflict; Fischer, “Electoral Conflict and Voting: a Strategy for Study and Prevention.” 30 Collier, Wars, Guns and Votes. 31 Höglund, “Electoral Violence in Conflict-Ridden Societies.” 32 Barkan and Ng’ethe, “Kenya Tries Again.” 33 “Briefing: Kenya’s Coastal Separatists—Menace or Martyrs?” 34 Ross, “Violent Democracy.” 35 Data on political violence incidents taken from the African Conflict Location and Event Dataset (ACLED). The data is coded so as to include violent events such as battles, riots, violent protests, and remote attacks perpetrated by groups such as governments, rebels, militias, ethnic groups, political organizations, and civilians. 36 ACLED. 37 ACLED. 38 Lindberg, “Democracy by Elections.” 22

2 The Political Context of Electoral Violence

Elections are meant to supplant other, often violent methods of political succession such as coups and assassinations. They are touted as a peaceful alternative to the political strife that frequently takes place in developing countries over access to political power. And yet, since the widespread adoption of multiparty elections in Africa in the early 1990s, elections more often result in violent contestation than they do in peaceful transfers of power. Election violence may be deliberately orchestrated or it may be unplanned. As the cases of Senegal and Liberia demonstrate, incidental violence can be sparked by perceptions of fraud or unfairness in the electoral process. Routine electoral events such as rallies turn deadly when aspirants and their supporters feel that outcomes are predetermined and that the election will not reflect the preferences of the citizenry. Strategic violence, on the other hand, responds to a different set of incentives. Although many politicians will insist that election violence is spontaneous, often times there is a logic behind its use. Strategic violence has been used to affect elections in several countries in Africa including Kenya, Zimbabwe, and Nigeria. Given that political actors operate within the constraints imposed by their own political environs, it is possible to tease out the domestic factors that are associated with electoral violence. The electoral context, specifically the openness of competition and the fairness of the process, affects how parties and voters respond to electoral outcomes. Institutions and environments that promote exclusion and restrict competition may be more likely to result in incidental forms of electoral violence. Additionally, as electoral institutions establish the formal means by which politicians access power, they also affect the viability of violence as an electoral strategy. This chapter will describe the nature of electoral vio23  

24

The Political Context of Electoral Violence

  lence in Africa and identify specific political factors associated with its occurrence. Describing and Explaining Electoral Violence

Partially due to the events in Kenya—but also due to violent elections in countries such as Angola, Democratic Republic of Congo, Ethiopia, Nigeria, Senegal, Togo, and Zimbabwe—electoral violence has emerged as a new field of study.1 Researchers have recognized that electoral violence, while related to political violence, is deserving of attention in its own right. Estimates of the prevalence of electoral violence in Africa vary widely. Because there are different operationalizations and measurements used to examine it, there are also different conclusions as to its prevalence. Some have found that it occurs in a majority of elections held in Africa. Straus and Taylor (2012) estimate that approximately 60 percent of all elections held in Africa between 1990 and 2008 resulted in some form of pre- or post-election intimidation or violence.2 According to Daxecker (2012), 81.48 percent of elections held between 1997 and 2009 in Africa experienced at least one pre-election conflict event and 68.8 percent experienced at least one post-election conflict event.3 Based on her data, the average number of pre-election conflict incidents is 23.4 per elections. The average number of post-election conflict incidents is slightly lower at 15.28. Goldsmith (2014), on the other hand, argues that much of the existing data on electoral violence in Africa dramatically overstates its occurrence. Using yet a different source of data, he finds that between 25 and 33 percent of elections held in Africa between 1990 and 2010 had violent incidents occur either before or after Election Day. However, Goldsmith isolates his analysis to only executive elections and excludes first round elections when two rounds take place to determine the victor. His defense is that these elections are the most competitive and most polarizing, thus most likely to exhibit electoral violence.4 Unfortunately, he provides no data to support this claim. In the data analysis section below, I demonstrate how his assumptions are incorrect and that the inclusion of parliamentary and first-round executive contests are necessary for estimating the frequency with which electoral violence takes place. Based on my own analysis, I believe the true frequency with which electoral violence occurs is closest to the numbers offered by Straus and Taylor: between 50 and 60 percent of elections in a given year in Africa experience some form of violence. However, as will be explained fur-

Electoral Violence in Sub-Saharan Africa 25

ther below, this number masks the different types and intensities of the violence that is taking place. Writing in 2009, Höglund was one of the first to explicitly define electoral violence as a unique subset of political violence.5 She identified several potential triggers for electoral violence including clientelism (especially the exclusionary nature of clientelistic relationships), underdeveloped political party systems, previously armed actors who are able to be mobilized with relative ease, and weak judicial and security sectors. These factors are frequently found in new democracies and countries in transition. Since Höglund’s initial call for a more nuanced analysis of election violence separate from that of political violence, there have been several potential reasons put forward to explain why electoral violence takes place. Unsurprisingly, because of the severity of the Kenyan case, several researchers have analyzed the 2007-2008 post-election violence in more depth. Derçon and Gutierrez-Romero (2012) conducted surveys both before the contentious 2007 Kenyan elections and after the postelection violence in 2008.6 Based on a comparison of survey responses, they conclude that Kenya’s post-election violence was a result of the widespread perception of electoral irregularities and the ineffectiveness of the Electoral Commission of Kenya (ECK) in resolving electoral disputes. Similarly, Optiz, Fjelde, and Höglund (2013) examine election management in Malawi, Ethiopia, and Zanzibar and find that electoral management body independence decreases the likelihood of violence taking place after an election.7 Electoral factors are not the only ones that have been proposed to explain the Kenyan case. Mueller (2008) points to Kenya’s weak political institutions and hyper-presidentialism as major causes.8 Still others believe that the Kenyan violence, specifically, was essentially a landbased conflict that the political elite manipulated and stoked to affect the election (Boone 2011; Rutten and Owuor 2009).9 These explanations, however, may not be generalizable to other cases as these researchers have largely focused their analysis, whether explicitly stated or not, on post-election violence. Pre-election violence presents as theoretically distinct in what it can and cannot affect due to the timing of its occurrence. Election violence that takes place after the vote is meant to subvert or alter an already determined outcome whereas election violence that takes place before the vote is meant to reduce competitiveness and distort voter preference. Turning to the literature on pre-election violence specifically, Höglund (2009), Straus and Taylor (2012), Hafner et al (2014) all argue that the closeness or competitiveness of elections increases the likeli-

26

The Political Context of Electoral Violence

  hood that electoral violence will take place. Derçon and GutierrezRomero, on the other hand, argue that the competitiveness of the Kenyan election was not a significant catalyst for electoral violence. Alleged electoral irregularities, however, may negate Derçon and Gutierrez-Romero’s finding as it suggests the electoral results were not a true reflection of political competition. The true level of electoral competition may have been much higher than the fraudulent results indicate. Obakhedo (2011), analyzing election violence in Nigeria, argues that increased civic education—with an emphasis on the nuances of election management—could help decrease the levels of election violence often found in Nigeria.10 His analysis, however, does not differentiate between types of electoral violence or the timing at which it occurs. Fjelde and Höglund (2014) examine how the specific electoral rules for translating votes into seats affect how often electoral violence takes place. They argue that elections held using majoritarian systems and smaller district magnitudes are more prone to electoral violence than elections held using proportional representation. They also find that the size of excluded minority groups and the level of land inequality in a given country can compound the effects of electoral rules. Fjelde and Höglund structure their analysis so as to measure both pre-election violence and post-election violence, but do not identify the factors that affect pre-election violence and post-election violence. While not necessarily an impediment to their overall conclusions, treating electoral violence as a monolithic occurrence obscures some of the more granular details that might help explain its causes. Several authors have also identified electoral fraud as a factor that affects the likelihood of electoral violence breaking out.11 Electoral fraud, or the belief that it has or will take place, motivates protesters to take to the streets and creates opportunities for violence. Fraud and violence are highly correlated. According to a global dataset compiled by Judith Kelley (2012), voter fraud occurs in approximately 47 percent of elections in which pre-election violence has taken place. Irregularities in the vote tallying process occur in 56 percent of elections in which preelection violence takes place.12 Voter fraud occurs in approximately 48 percent of elections in which voter intimidation occurs. Irregularities in the vote tallying process occur in 60 percent of elections in which voter intimidation takes place. Electoral fraud may provide a partial explanation for incidental electoral violence, but it is unable to fully explain strategically motivated violence. Additionally, there is a temporal component that structures the relationship between fraud and violence. Planning for electoral fraud

Electoral Violence in Sub-Saharan Africa 27

may take place well in advance of the election but activities such as ballot box theft, ballot box stuffing and manipulated tallies occur on the day of elections. And election days tend to be comparatively free of violence. According to Kelley’s data, violence on Election Day itself occurred in less than 30 percent of elections held between 1980 and 2004. In countries in which no pre-election violence took place, only 14 percent experienced violence on Election Day. In sum, while there has been a recent concerted effort to describe and identify patterns associated with the occurrence of electoral violence, there has been no consensus on how often it takes place and what exactly causes it to occur. Nowhere is this more apparent than with the complex and sometimes contradictory explanations for Kenya’s 2007 2008 crisis, which include competitiveness, fraud, a centralized presidency, democratic degradation, and land disputes. Furthermore, while Kenya is a very important case to understand, the 2007 - 2008 election violence is a rather unique in that the majority of the violence and fatalities occurred after the election. Post-election violence differs from preelection violence in terms of timing, method, and intent. It is disruptive and it is generally a reaction to the electoral outcome, not a strategy to affect voter or aspirant behavior before Election Day. It may be meant to bring attention to electoral irregularities or may be used by losing candidates as a way to leverage themselves into government after a vote. As such, the causes of the Kenyan post-election violence or other such high-profile crises may not be directly or fully germane to cases of preelection violence. In sub-Saharan Africa, approximately 95 percent of violent elections experience pre-election violence, 24 percent of violent elections experience both pre and post-election violence, but less than 4percent of violent elections are exclusively characterized by post-election violence.13 Although it is typically more fatal and more intense—between 1,300 and 1,500 fatalities after the 2007 Kenyan election, more than 800 fatalities after the 2011 Nigerian election, and several thousand dead after the 2010 Ivoirian election—post-election violence occurs much less frequently than pre-election violence, which suggests that pre-election violence deserves its own examination. Because of the predominant number of high-quality analyses of post-election violence, we know somewhat more about how and why it occurs. We know substantially less about the nature of pre-election violence. Furthermore, there is evidence to suggest that pre-election violence is a significant predictor of post-election violence. In the Kenyan PEV, for example, districts that were affected by pre-election violence were almost 50 percent more likely to experience post-election violence.14 Straus and Taylor (2012) also conclude

28

The Political Context of Electoral Violence

  that it is rare for post-election violence to take place without pre-election violence taking place first.15 The remainder of this chapter is devoted to the examination of pre-election violence. Motivations and Perpetrators of Pre-Election Violence

Pre-electoral violence is generally either strategic or incidental. Incidental electoral violence takes place when highly emotional political actors come in close proximity with other actors in opposition prior to an election. Protests, rallies, and the like may turn violent as an expression of frustration over some unfairness, real or perceived, in the electoral process. They key distinction between incidental and strategic violence is that incidental violence is not planned in advance. Its spontaneity, however, should not mask the fact that political and institutional factors indirectly affect its prevalence as these can affect the perception of the fairness and unfairness of the electoral process. Nonetheless, it is difficult to ascertain exactly when and how it will break out due to lack of explicit planning or mobilizing. Strategic electoral violence, while perpetrated by any number of actors, is encouraged or mobilized at the elite level and the state is frequently complicit. According to Straus and Taylor (2012) and Fjelde and Höglund (2014), political incumbents most often initiate preelectoral violence. Governments use state security forces, party youth wings, or hire militia groups to act on their behalf. The electoral violence in Zimbabwe in 2000, 2002, and 2008 was largely orchestrated by the ruling ZANU-PF (Zimbabwe African National Union-Patriotic Front) and perpetrated by youth militias that the party assembled and employed.16 Similar scenarios have played out in Kenya and Nigeria. Electoral violence in Kenya in 1992 and 1997 was a political strategy of incumbent Daniel arap Moi and KANU party elites, who hired several different groups including Kalenjin warriors to intimidate, kill, and displace voters.17 Incumbents, however, are not the only actors mobilizing violence. Before the 1992 elections, both KANU and the opposition Forum for the Restoration of Democracy (FORD) reportedly created youth wings for the express purpose of intimidating and harassing candidates and their supporters.18 During the 2002 and 2007 Kenyan elections, the Mungiki (a quasi-religious criminal organization) were believed to participate in pre-election intimidation on behalf of the opposition.19 Political aspirants in Nigeria also employ violence as an electoral tactic. Politicians from the ruling party and opposition political parties routinely hire gangs as part of an overall campaign strategy that includes both vote

Electoral Violence in Sub-Saharan Africa 29

buying and violence.20 In Tanzania, the ruling party and main opposition party have both organized and trained illegal militias to assist in their informal campaigning.21 The people who have the most to gain (or lose) from strategic preelectoral violence are political and electoral actors. Therefore, it is instructive to examine in the abstract why these individual actors might choose to mobilize electoral violence. The purpose of this exercise is to uncover the logic of electoral violence from the perspective of political elites—the primary decision makers in the electoral violence calculus— and use this to identify what institutions may encourage or discourage the strategic use of electoral violence. One of the underlying assumptions of this analysis is that politicians are rational office seekers.22 Their ultimate goal is to compete for and win elected office. Ideology, policy, and political parties are all just a means to this end. Aspirants must have some a priori knowledge and beliefs about their political environs, i.e., who previously won office, by what margin, who their likely challengers will be, how competitive the race will be, and who their base of support will be, etc. Prior to entry, politicians make a quick estimate of their chances of success in a given political race. Many in high-information environments go so far as to conduct polls or assemble focus groups in order to ascertain the level of electoral support and likelihood of victory. In the African case, electoral polls are routinely conducted in Kenya, Nigeria, and South Africa, amongst other countries. After the decision to enter a race has been made, an individual aspirant has the ability to estimate their electoral fortunes at any given time prior to an election. They do this despite having perfect and/or complete information about the electoral landscape. A few months before an election is scheduled a political aspirant is able to surmise how much support they have and how much they may need in order to win the election. With this estimate, the aspirant can then examine the array of options they have to influence their chances of winning. Regardless of the method, the goal is to reduce the competitiveness of the race, either by increasing one’s own vote share or decreasing the vote share of those in opposition. If the incumbent has a reasonable estimate of the “costs” of violent strategies and the “benefits” of elective office, they can do some simple calculations to make reasonable guesses as to what the impact of both violent and nonviolent strategies may have on their chance of winning office. With the assumption still that the incumbent wants to maximize their chances of success, they then will choose the option associated with the highest payoff—the winning of office.

30

The Political Context of Electoral Violence

  One would expect electoral violence to occur when the benefits significantly outweigh its costs. Admittedly, the assumption that political aspirants are agnostic about the use of violence to achieve their goals may be unrealistic, but the fact that significant numbers of politicians have proven their indifference to the use of violence suggests it is an assumption worth exploring. The prevalence of violence alone suggests that it is an accepted strategy to many politicians. By analyzing how political elites could assess potential benefits and costs of employing violent election strategies, it is possible identify what specific political institutions affect the final calculus. The Role of Electoral Institutions

What this simplified analysis assumes is the costs of violent strategies and the benefits of winning office are knowable. They may not be, but to derive a reasonable estimate one can analyze what factors could increase or decrease costs and benefits and to what comparative degree. There may also be factors that affect the ability of an aspirant to gauge the distribution of electoral support and her likelihood of winning office. Electoral institutions establish the rules for how voter preferences are translated into representative governments. Electoral institutions also explicitly define success, establish the minimum threshold for success and affect the ability of political aspirants to predict their likelihood of success. As such, different institutions incentivize different strategic behaviors. Additionally, because electoral institutions can create such varied outcomes in terms of the number of parties that participate in elections23 and the overall representativeness24 of elected bodies, elites often disagree about the most appropriate types of electoral systems given their relative strengths and overall objectives. Elites choose specific electoral institutions (and change them as they see fit) based on their own political objectives. For instance, opposition parties often demand that majoritarian rules or proportional electoral rules be used because they believe them to produce fairer outcomes. Proportional representation, specifically, provides smaller and weaker parties more opportunities to win office, which is why opposition parties have a strong preference for these rules in particular. Parties in power, on the other hand, typically prefer plurality systems—even though the minimum threshold for victory may be lower at the constituency or district level—because to succeed in national elections that use plurality rules typically requires a national presence and the ability to campaign across the country. This is difficult for smaller parties and parties with

Electoral Violence in Sub-Saharan Africa 31

limited resources. Additionally, in postconflict settings such as South Africa, Rwanda, and Burundi, where politics have been especially polarized, proportional representation is often times proposed to encourage co-operation and power-sharing. The ability of elites to choose the particular electoral institutions could feasibly create an endogeneity problem with this analysis. Perhaps it is not the institutions themselves that affect the likelihood of violence but rather it is unscrupulous political elites who chose specific institutions that are more prone to manipulation. Or maybe because certain institutions are proposed in postconflict settings, these settings are less likely to experience electoral violence, not because of the institutions but because of the impact of previous conflict. The problem with this argument, however, is that it overlooks the transformative power of political institutions and the ways in which rules in particular channel behavior. It also suggests that violent societies (or those recovering from intense conflict) may choose institutions that are less prone to violence whereas countries with no such history may deliberately choose institutions that promote violence. It does not seem likely, however, that elites in countries with no history of violent conflict may find themselves choosing violence-prone institutions. There are three primary types of electoral systems with much variation in between: plurality and majoritarian systems, proportional representation systems, and mixed systems, which combine aspects of the two. Plurality and majority electoral rules assign winners and losers for each individual seat or political race using single-member districts whereas proportional representation assigns a percentage of seats to parties based on their percentage of vote support. Proportional representation requires multi-member districts, which vary in size but can in theory be as large as the number of seats available. Elections held using plurality and majoritarian systems are sometimes referred to as “winner-take-all” races. There are obvious winners and losers and the minimum threshold can be very low. In Zambia, the president is elected according to plurality rules. The candidate with the most votes wins. Levy Mwanawasa won the 2001 Zambian presidential election with 29.15 percent of the vote. The remaining 70 percent of the vote was split amongst 10 other candidates. The runner-up received 27.2 percent of the vote, less than 2 percentage votes less than Mwanawasa. Similarly, Moi won the 1992 Kenyan presidential election with 36.35 percent of the vote to Kenneth Matiba’s 26 percent. Because of the possibility of an extreme imbalance between voter support and the winner of an election, some have argued that simple plurality electoral rules are inherently unfair and encourage divisive pol-

32

The Political Context of Electoral Violence

  itics.25 For presidential elections, which by nature require that one person alone wins, some countries have adopted majoritarian or qualified electoral rules as an attempt to engineer a fairer and more representative outcome. Under these rules, the ultimate winner must receive support over and above a certain threshold or a second round of voting is required in which voters choose between the two top vote getters from the first round. This is meant to encourage candidates to assemble a larger coalition of support and ideally one that is more representative of the population. Kenya explicitly adopted a qualified plurality system after its experiences with Moi’s simple plurality victories in the 1990s in which his electoral support came from specific geographic regions and ethnic groups. For the 2002 and 2007 elections, the winner of the presidency was required to receive at least 25 percent of the vote in five of the country’s seven provinces, or a second-round of voting would take place. For the 2013 election, this requirement was increased to over 50 percent of the vote and at least 25 percent support in at least half of the country’s 47 counties. In 1999, Nigeria also adopted measures to ensure that its political parties must assemble a more representative base of support and stop relying on existing ethnic, regional, and religious cleavages. To this end, successful presidential candidates must win a plurality of votes at the national level and at least 25 percent of the vote in two-thirds of the country’s 36 states. Because they are multi-member bodies, legislatures can be apportioned based on plurality rules, majoritarian rules, proportional representation rules, or some combination thereof. Proportional representation awards seats based on the percentage of electoral support a political party receives. By its very nature, proportional representation cannot be applied to elections that produce a single winner (i.e., executive elections). Whereas plurality and majoritarian systems produce a single winner from each district or constituency, proportional representation requires multi-member districts. As its name would suggest, the electoral outcomes produced by this system are inherently more proportional and many have argued for its adoption in Africa’s new democracies. This has been especially true for heterogeneous countries and those overcoming deep or polarized social divisions, such as found in South Africa and Rwanda.26 Depending on the specific formula, threshold, and district magnitude, even very small parties with limited electoral support can win office in systems that use proportional representation, creating the opportunity for more inclusive governments. Members of the Beninese National Assembly, for example, are elected using a system of closed-

Electoral Violence in Sub-Saharan Africa 33

list proportional representation, meaning that party leadership is responsible for presenting a list of candidates to voters and voters cast their ballots for the party lists, not individual candidates. Seats are then awarded to the parties based on the proportion of the total vote that they receive and the number of seats available in each multimember constituency. The 2011 election resulted in almost 1,600 candidates from 19 party lists competing for 83 seats. Eight parties were awarded seats with the two largest parties sharing 71 seats and the remaining six parties receiving two seats each. Each electoral system has its own merits in terms of efficiency, decisiveness, accountability, representation, and fairness. Some argue that plurality and majoritarian systems, because they produce a single victor, foster stronger linkages between politicians and citizens by forcing politicians to remain accountable to a specific constituency or district. Others argue that proportional representation encourages more consensual politics and reduces conflict at large due to the fact that it creates fairer electoral outcomes. Because of the various arguments in favor and against these systems, some countries have opted to combine them to create hybrid or mixed electoral systems in an attempt to exploit the advantages of each system. Lesotho, for example, replaced its plurality electoral system with a mixed system prior to its 2002 elections. The previous elections in 1998, in which the ruling party won 79 out of 80 seats (98.75 percent of seats) despite receiving only 60.5 percent of the total vote, led to massive unrest and concern that a coup was imminent. The opposition—frustrated that it was being essentially excluded from parliament—protested, rioted, and burned down part of the capital city.27 The Southern Africa Development Community (SADC) was forced to intervene and negotiated a settlement between the ruling party and the opposition that included the adoption of a new electoral system intended to promote a more representative government. Under Lesotho’s mixed member system, 80 of the 120 available seats are now filled through plurality rules in single-member constituencies. The remaining 40 seats are awarded based on proportional representation using one nationwide constituency. Under proportional representation, the meaning of electoral success/failure is not the same as under plurality rules and the ability to predict electoral outcomes is more difficult, especially when district magnitude, or the number of seats to be filled, varies across constituencies. As these types of electoral institutions encourage different types of electoral strategies we should expect to find different outcomes in terms of the use of violence. Elections using plurality/majoritarian rules be-

34

The Political Context of Electoral Violence

  come do or die affairs since there are clear winners and losers. Elections using proportional representation, instead of sorting individual candidates into winners and losers, assign victories of magnitude in terms of representation. Also, because it is easier to predict and therefore presumably affect elections under plurality electoral systems, these types of elections should be more prone to electoral violence. Additionally, plurality systems have a low minimum threshold for victory (one vote more than everyone else), specifically in comparison to majoritarian systems, so the use of violence should be more effective at reducing competitive margins and producing electoral victory. However, because of the relationship between regimes and electoral management, it could be the case that the effects of electoral institutions are muted under more repressive or autocratic environs. The expectations that we have regarding how electoral institutions structure behavior implicitly assume high-information environments and unbiased institutions. Mylonas and Roussias (2007) argue that electoral rules adhere to our common expectations in more democratic countries. In nondemocracies, the ability of the government to exert influence over various electoral processes (i.e., appointment of electoral management body officials, engaging in vote rigging with less chance of detection, and employing state resources during campaigns) all work to reduce the impact of electoral rules on party and voter behavior.28 The subsequent analysis explicitly takes this possibility into account. Costs of Electoral Violence

The costs of electoral violence can be understood in terms beyond injuries and fatalities and the loss of property. These categories of cost are sometimes incurred by political elites, as in cases of political assassinations, but much of what constitute the costs of electoral violence to political elites and decision-makers has to do with getting caught and punished—either electorally or judicially. From a campaigning perspective, environments in which politicians can act with relative impunity make the employ of electoral violence less costly than those in which there are enforced penalties (fines, imprisonment) for inciting violence. Unfortunately, in many African countries domestic political and penal institutions do not have the mandate or capacity to effectively monitor and punish offenders. Independent electoral management bodies and independent and effective judiciaries increase the costs of employing electoral violence. An independent and professional media also increases the cost of violence as it increases the chances of being reported

Electoral Violence in Sub-Saharan Africa 35

and/or caught. Overall, strong and independent political and penal/judicial institutions increase the costs associated with electoral violence through their ability to detect instances of electoral violence and punish offenders. Effective monitoring, reporting, and punishment mechanisms all deter violence. Impunity increases the likelihood that electoral violence will occur. In sum, the features of substantive democracy—protections for civil liberties, press freedoms, and effective checks and balances on political power—all promote the integrity of elections, one of the key procedural components of democracy, and reduce the likelihood that electoral violence will occur. Because of the complicated relationship between elections and democracy—democracy requires elections but elections do not require democracy—this is a subject that I will return to throughout this book. Data Analysis

In the simplest cost-benefit analysis, in order for electoral violence to be a worthwhile strategy, the benefits of engaging in electoral violence must outweigh the costs. The above analysis makes clear that there are important institutional factors to examine in order to better understand why politicians may choose to incite and/or mobilize electoral violence. There are also institutional factors that promote fairer and less fair outcomes, which indirectly affect how individuals interpret electoral results. From a theoretical perspective, we can identify factors that may increase or decrease the likelihood of electoral violence taking place, but until recently no systematic attempt to measure electoral violence existed, making empirical analysis impossible. While there are numerous examples of attempts to quantify and examine political violence (Africa Conflict Location and Event Dataset, Major Episodes of Political Violence, Political Terror Scale, Social Conflict in Africa Database, etc.), analyses of electoral violence have relied primarily on case studies restricted to a few high profile cases of post-election violence such as Kenya and Côte d’Ivoire. The research conducted has provided us with invaluable case-specific insights but it remains to be seen if there are generalizations we can make about electoral violence that transcend time and place, especially in light of the fact that the majority of electoral violence that takes place in sub-Saharan Africa occurs prior to the election and not afterwards. Straus and Taylor (2012) have assembled a dataset, the African Electoral Violence Database (AEVD), which captures data on the level of violence occurring in every African election that was held between

36

The Political Context of Electoral Violence

  1990 and 2008. Using previous data collected by Staffan Lindberg29 supplemented by State Department human rights records, they assess the level of violence that took place in every election held in sub-Saharan Africa between 1990 and 2008. They analyze a total of 221 elections and assign a numerical value ranging from 0 to 3 to each election: 0 for elections that experienced no violence; 1 for those that experienced violent harassment30; 2 for those that experienced violent repression, a category including political assassinations and prolonged imprisonment of political actors; and 3 for elections that experienced generalized and widespread violence, indicated by 20 or more fatalities. Using their operationalization and data collection method, I was able to update their dataset to include elections that took place through the June 2014. Although admittedly a rather blunt way of capturing the differences between my conceptual categories of strategic and incidental electoral violence, categories 2 and 3 (violent repression and generalized widespread violence) most resemble what I term “strategic violence” as these categories include assassinations, imprisonment, forced displacement, and widespread violent action. Category 1, violent harassment, includes “police or security forces breaking up rallies, party supporters fighting, street brawls, opposition newspapers being confiscated, candidate disqualifications, and limited short-term arrests of political opponents.”31 While violent harassment certainly could encompass strategic behaviors such as candidate disqualifications and the like, this category is closer to “incidental” episodes of electoral violence, especially as some of its included components are forms of electoral fraud and not explicitly violent. The distribution of pre-election violence across elections in Africa is found in Figure 2.1. Of the 289 elections that took place in Africa between 1990 and 2014, 43 percent (124 elections) experienced no pre-election violence while the majority, 57 percent (165 elections) experienced some form of violence and/or intimidation. The most common form of electoral violence is what Straus and Taylor define as “violent harassment” and I term “incidental.” One hundred and six elections (37 percent) held during this time period experienced violent harassment. Thirty-two elections (11 percent) experienced violent repression and twenty-seven elections (9 percent) experienced large-scale violence. There is also variation between countries that routinely experience pre-electoral violence (Côte d’ Ivoire, Democratic Republic of Congo, Kenya, Nigeria, Sudan, Togo), those in which pre-election violence never occurs (Botswana, Cape Verde, Mauritius, São Tomé, and Príncipe, and Seychelles),

Electoral Violence in Sub-Saharan Africa 37

and those in which pre-election violence sometimes occurs (Benin, Burundi, Cameroon, Liberia, Uganda, and Tanzania). Figure 2.1: Incidences of Pre-Electoral Violence in Sub-Saharan Africa, 1990 to 2014

Large-Scale Violence 9% Repression 11%

Harassment 37%

No Violence 43%

n=289

Using an updated version of the AEVD dataset, it is possible to examine if the occurrence of pre-electoral violence is indeed affected by institutional constraints such as type of election, timing of elections, electoral system, and democratic quality. In addition the African Electoral Violence Database, data come from multiple sources including POLITY32 (which measures democratic/autocratic quality), Freedom House33 (which measures political freedoms and civil liberties), and the Mo Ibrahim Index of African Governance (which measures several indicators of good governance including participation, human rights, human development, safety, and rule of law).34 The following section examines the implications from the above pre-electoral violence calculus and compares each to the empirical record. Political systems that concentrate executive power in the hands of an individual office holder (president or prime minister) create significant incentives for the winning of this specific office. Few systems in Africa (or the world) have chosen to adopt a multiple executive system (South Africa during the transition from apartheid to multiparty elections in the 1990s is one example). Parliamentary systems share executive power between an executive and the legislative branch but these are also rare in Africa (Botswana and South Africa are two examples). Intuitively, electoral institutions that increase or decrease the “benefit” or

38

The Political Context of Electoral Violence

  “reward” of political office should increase the pay-off of wining office, increasing the likelihood of both incidental and strategic violence occurring. Plurality systems, also commonly referred to as winner-take-all systems, reward individual office-seekers as opposed to political parties but they also concentrate risks and costs as opposed to parties, which can share costs and risks. Additionally, there are few incentives to build larger coalitions because the individual with the most support, not a majority of support, wins. All other entrants lose. Majoritarian electoral systems require that executive candidates secure a majority of votes either through “natural” bases of support or through a manufactured majority by mandating a second round run-off system, pitting the two highest vote getters against each other in a follow up election. As such, plurality systems should create more incentives for strategic violence than majoritarian electoral rules (50 percent plus 1). Figure 2.2: Pre-Electoral Violence and Electoral Rules, Executive Elections

100

Percent

75 50

Strategic Violence Incidental Violence

25

No Violence

0 Plurality Rules Majoritarian Rules (n=50) (n=117) Executive Rules

Figure 2.2 reports the differences in incidences of electoral violence (none, incidental, and strategic) between plurality and majoritarian executive electoral rules. In sub-Saharan Africa, plurality executive elections are more prone to incidences of pre-electoral violence in general; 77 percent of plurality elections experience some form of pre-electoral violence (incidental or strategic) whereas 54 percent of majoritarian

Electoral Violence in Sub-Saharan Africa 39

elections experience some form of pre-electoral violence. Alternatively stated, 22.4 percent of plurality executive elections experienced no form of pre-electoral violence compared to 46.5 percent of majoritarian elections. When analyzed by type of electoral violence, plurality rules again appear to be associated with higher frequencies of incidental violence and strategic violence, with strategic violence occurring the most frequently. Legislative elections, in general, may be less prone to violence than executive elections; however, the same relationship between electoral system and incidence of pre-electoral violence should hold. The added caveat to this relationship would be the effect of proportional representation electoral rules on the incentives and payoffs for violence. Proportional representation rewards seats based on the percentage or proportion of the vote a party receives. It is a method of elected power sharing and should create fewer incentives for violence, as it does not completely block a party, or its office seekers, from accessing political power. Rather, proportional representation rewards seats and legislative strength based on their percentage of voter support. For instance, if a legislature is comprised of 100 seats and Party A receives 75 percent of the vote and Party B receives 20 percent of the vote and Party C receives 5 percent of the vote, Party A would be awarded 75 seats, Party B 20 seats, and Party C 5 seats. Figure 2.3: Pre-Electoral Violence and Electoral Rules, Legislative Elections

Percent

100 75 50 25

Strategic

0

Incidental No Violence

Legisla(ve  Rules  

40

The Political Context of Electoral Violence

  As shown in Figure 2.3, legislative elections that use proportional representation rules are less prone to both forms of pre-electoral violence than are plurality elections. Approximately 53 percent of elections using proportional representation are peaceful compared to 42.9 percent of majoritarian elections and 27.6 percent of plurality legislative elections. Mixed electoral systems—those that combine features of plurality rules with those of proportional representation—most resemble proportional representation elections in terms of frequency of violent elections as 56.7 percent of elections held under mixed systems experience no preelection violence. Incidental electoral violence takes place more frequently in both plurality and majoritarian electoral systems—both of which are argued to be inherently less fair types of electoral rules. Strategic electoral violence most frequently occurs under plurality electoral rules (29.7 percent) compared to all other electoral rules, which range from roughly 13 to 15 percent. Plurality rules, specifically, have been argued to lend themselves to the types of machinations that strategic electoral violence seems to inspire. While it is notable that legislative plurality elections are marginally more violent than executive plurality elections, this examination of electoral system and level of pre-electoral violence does not control for timing of elections. The next section examines the type and timing of elections and frequency of pre-electoral violence. Nonetheless, there is support for the notion that different electoral systems provide different incentives for the use of pre-electoral violence and confirms some of the relationships identified by Fjelde and Höglund (2014). The timing/type of elections also affects the benefit of holding elected office. Concurrent elections, in which presidential and legislative offices are presented to the electorate at the same time, offer the opportunity for a well-organized and coordinated party to win all possible elected offices at the same time and, hence, control almost the entire state apparatus. Because legislatures diffuse political power, the benefit of winning legislative office (while great) is still lesser than the benefit of winning executive office. This is especially the case in sub-Saharan Africa where legislatures are generally weak35 and executives are excessively powerful, an arrangement sometimes referred to as hyperpresidentialism.36 Figure 2.4 reports the frequency of pre-electoral violence by type and timing of election. Legislative elections are more often peaceful than executive or concurrent elections—55 percent of legislative elections are peaceful compared to 37.3 percent and 34.3 percent for execu-

Electoral Violence in Sub-Saharan Africa 41

tive and concurrent elections, respectively. While this relationship is significant, this does not mean that legislative elections should be considered peaceful affairs. Approximately 45 percent of legislative elections experience some form of pre-election violence. Despite what Goldsmith (2014) has asserted—that legislative elections are inherently more peaceful and thus should be excluded from analysis of electoral violence—these elections still exhibit significant levels of electoral violence and thus must be included in any discussion of the prevalence of electoral violence in Africa. As expected, executive elections and concurrent elections exhibit similar levels of pre-electoral violence. From this analysis it is impossible to say whether this is because of the presence of executive elections (common to both) or some other unmeasured factor. Figure 2.4: Pre-Electoral Violence and Type/Timing of Elections

Percent

100 75 50 Strategic Violence

25

Incidental Violence

0 Legislative Executive Concurrent Elections Elections Elections (only) (only) (both) (n=114) (n=75) (n=99)

No Violence

Type/Timing of Election

As discussed previously, strong institutions, including judicial and penal institutions, can increase the costs of engaging in electoral violence. Electoral management bodies that monitor elections for irregularities and perform impartial dispute resolution functions increase the costs of electoral violence. Effective and independent judiciaries that have the political will and capacity to try (and convict) perpetrators of electoral violence also greatly increase the costs associated with electoral violence. Unfortunately, there is no cross-national data on judicial effectiveness or the strength and autonomy of electoral management bodies. One way of examining this relationship would be to look to the broader

42

The Political Context of Electoral Violence

  quality of democracy in a given country. There are many ways of measuring quality of democracy but Polity and Freedom House scores are two of the most frequently used. Polity assigns annual scores ranging from -10 (hereditary monarchy) to 10 (consolidated democracy) for all countries beginning in 1800. Polity aggregates data on institutional rules regarding executive recruitment, executive constraints, political participation, and political competition in order to provide a yearly measure of autocratic/democratic quality. It is meant to measure institutional restraints/constraints on elected leaders and voters. It stands to reason that the higher a Polity score/more democratic a country, the more effective are the political institutions in a given country. Figure 2.5: Pre-Electoral Violence and Democracy (Binary)

Percent

75

50 No Violence

25

Incidental Violence Strategic Violence

0 Democratic Spectrum (n=156)

Autocratic Spectrum (n=133)

Regime Type

Figure 2.5 reports the relationship between frequency of pre-electoral violence and quality of democracy in Africa using Polity Scores. For ease of analysis, Polity scores are divided into two categories: those on the autocratic spectrum (ranging from -10 to 0) and those on the democratic spectrum (ranging from 1 to 10). These two categories of democratic quality exhibit statistically different incidences of electoral violence. There are more than twice as many peaceful elections held under systems on the democratic end of the spectrum (59 percent) as there are held under the autocratic end of the spectrum (24.1 percent). Additionally, strategic violence more frequently takes place under autocratic regimes than those that fall on the more demo-

Electoral Violence in Sub-Saharan Africa 43

cratic end of the Polity spectrum. It seems likely that strategic violence may be more difficult to utilize under countries with more open and competitive political institutions. This dichotomous treatment of democracy, however, might be masking important differences between more consolidated regimes and those in transition. Some researchers examine democracy using three distinct categories: autocracies, anocracies (or those countries in transition), and democracies. Polity scores can be subdivided to capture these three categories. When examined in this way, the differences between incidences of violence and democratic development become even starker (See Figure 2.6). Autocratic elections experience significantly more violence than either democratic regimes or those considered to be in transition. Of the 46 elections held by autocratic regimes, 87 percent experienced some form of violence, the majority type (56 percent) being incidental. This finding may reflect a willingness on the part of state security forces to use a heavy-hand in dealing with the opposition. Figure 2.6: Pre-Electoral Violence and Democracy

100

Percent

75 50

Strategic Violence

25

Incidental Violence No Violence

0 Democratic Transitional Autocratic (n=107) (n=136) (n=46) Regime Type

Finally, yet another measure that captures some of the weaknesses of monitoring and enforcement of electoral infractions is the level of safety and adherence to the rule of law in a country. Data on this comes from the Mo Ibrahim index of African Governance (IIAG), which takes into account the judicial process and independence of the judiciary, accountability of public officials, and reliability of police services, amongst other factors, in assigning a score for each country. Scores range from 0 to

44

The Political Context of Electoral Violence

  100 where 0 is no rule of law and 100 is high adherence to the rule of law. These scores are only available for Africa and beginning in 2000. Again, the pattern is reinforced: electoral violence is more common in countries with lower average rule of law and lower maximum rule of law than in countries that rank higher in rule of law (See Table 2.1). Countries in which electoral violence does not occur score, on average, 10 points higher in the categories of “Rule of Law” and “Accountability” than countries that experience any form of electoral violence. Additionally, those countries that experienced strategic violence) do not score any higher than 45.5 on the 100-point scale for accountability, whereas countries in which no violence or only violent harassment takes place have received scores in the upper 70s. One interpretation would be that low adherence to rule of law, especially in terms of accountability, is a necessary but insufficient condition for electoral violence to occur. Perhaps it must be paired with other election-specific factors such as competitiveness of the election to sufficiently motivate election-related violence. Table 2.1: Pre-Electoral Violence and Rule of Law & Accountability

No Violence

Incidental Violence

Strategic Violence

n

Rule of Law Average

Accountability Average

(Mo Ibrahim Index)

(Mo Ibrahim Index)

56.9

44.8

(77)

(77)

47.3

37.9

(64)

(64)

41.7

31.3

(28)

(28)

169

169

In order examine how both political institutions and the political environment affect the likelihood that pre-electoral violence takes place, it is necessary to employ inferential statistics. Ordered logistic regression is the most appropriate statistical method given the ordinal nature of the dependent variable pre-electoral violence. This model allows predictions to be made as to the likelihood that an election will fall into one of the three categories of electoral violence (none, incidental, or strategic).

Electoral Violence in Sub-Saharan Africa 45

The key independent variables are electoral system type (plurality, majoritarian, mixed, proportional), type of election (singular or concurrent), democratic quality (measured by Polity score) and the interaction of democratic quality and electoral system. Electoral system is measured by a series of dummy variables so as to allow each type of rule to exert its own effect. GDP per capita is also included in the model as a control variable as it seems plausible that economic considerations factor into campaign decisions. Separate models were run for executive and legislative elections because one of the key independent variables—electoral system type—can only be linked to electoral violence in instances when those specific rules are used. For instance, plurality electoral rules for executive elections cannot be theoretically linked to electoral violence in legislative elections when only executive elections take place, and vice versa. Polity and the Mo Ibrahim Index of African Governance (IIAG), which attempt to capture similar concepts, are highly correlated and cannot be used in the same model. Polity scores also cover the entire time period under consideration whereas IIAG scores only begin in the 2000s. Thus, Polity scores were used in the reported model but the results still hold when other measures of democracy are used. Freedom House scores are frequently used as proxy measures for democracy but because they often include subjective assessments of the freeness and fairness of elections and explicitly include electoral violence as a consideration in their assignment of scores they must be used and interpreted very carefully. I conducted the same analysis using one-year lags of Freedom House scores so as to theoretically omit assessments of the quality and conduct of elections. The results are similar throughout. Additionally, using Polity scores allow us to move beyond the dichotomous treatment of “democratic spectrum” versus “autocratic spectrum” reported above. Table 2.2 reports the coefficients and standard errors of the ordered logit models estimating the probability that electoral violence takes place in executive elections and legislative elections. Executive elections held under plurality rules are associated with higher levels of electoral violence, especially strategic violence. Democratic quality exerts a significant effect on the likelihood of electoral violence taking place. Countries that rank higher on the Polity scale, meaning electoral competition and participation are relatively open, experiences less electoral violence.

46

The Political Context of Electoral Violence

 

Table 2.2: Pre-Electoral Violence and Executive Elections, Ordered Logit Regression Electoral Violence in Executive and Concurrent Elections Plurality Rules

.86**

(dummy)

(.39)

Democratic Quality

-.08* (.05)

Plurality Rules Interacted with Democratic Quality Type of Election

-.13* (.07) .62 (.39)

Executive Incumbent

.72** (.4)

GDP per capita

-.00002

n

127

Log-Likelihood

-119.40

Pseudo R2

.11

The interaction of electoral system and democratic quality allows for a more nuanced understanding of how these two factors operate together to affect electoral violence. Strategic violence is much more common under non-democratic regimes, but plurality systems are much more frequently associated with strategic violence than majoritarian systems. Plurality systems allow victors to emerge with potentially low levels of overall voter support as long as their level of support is com-

Electoral Violence in Sub-Saharan Africa 47

paratively higher than all other competitors. The threshold for victory can be very low and the payoffs for violence high. By design, majoritarian electoral systems require consensus building and the cobbling together or a majority of voter support. This may be more difficult to achieve through violence. Table 2.3- Pre-Electoral Violence and Legislative Elections, Ordered Logit Regression Electoral Violence in Legislative and Concurrent Elections Plurality Rules (dummy)

.62 (.41)

Majoritarian Rules (dummy)

(.63)

-.25

Mixed Rules (dummy)

.19 (.41)

Democratic Quality

-.18*** (.05)

Plurality Rules X Democratic Quality

-.02 (.08)

Majoritarian Rules X Democratic Quality

.28**

Mixed Rules X Democratic Quality

.04 (.07)

Type of Election

.61***

(.13)

(.19) GDP per capita

-5.9e-7

n

162

Log-Likelihood

-148.91

Pseudo R2

.12

48

The Political Context of Electoral Violence

  The presence of an executive incumbent participating in an election is significantly and positively related to the outbreak of election violence. This might reflect the access that incumbent presidents have to security forces or the intense desire of executive presidents to stay in power once they are in office. Interestingly, GDP per capita does not seem to have much effect on the likelihood of electoral violence taking place in executive elections. The relationship between electoral institutions, quality of democracy and electoral violence in legislative elections is more complicated. Table 2.3 reports the results of the ordered logit model estimating pre-election violence in legislative elections. A similar model is estimated for legislative elections but the presence of an incumbent running becomes more difficult to measure under legislative elections due to the fluidity of parties in Africa. Party switching is a common occurrence and parties are often viewed as vehicles for individual candidates.37 As such, it is very difficult to define and identify incumbent parties. Democratic quality, again measured through Polity scores, is negatively and significantly related to the outbreak of electoral violence. More democratic regimes are less likely to experience electoral violence, even when electoral rules are controlled for. When electoral institutions are interacted with democratic quality, electoral institutions become a much less significant and inconsistent factor in explaining electoral violence. The differences between plurality, mixed and proportional representation rules all seem to be largely unaffected by democratic quality, as all three are associated with similar levels and types of violence in nondemocratic regimes. Nondemocratic elections using any of these three types of rules are generally more violent than more democratic elections. Majoritarian electoral rules, however, seem to be more peaceful when held under nondemocratic regimes when held under more democratic regimes. This finding is most likely a function of limited data as very few countries use majoritarian legislative rules. Only 18 elections out of the sample of 289 use majoritarian rules for legislative elections. Type/timing of election is also related to incidences of electoral violence, meaning that concurrent elections are more likely to experience both types of electoral violence than are singular legislative elections. In order to get a better understanding of the persistence of electoral violence, it is important to examine its occurrence over time. Figure 2.7 charts the percentage of elections held in sub-Saharan Africa from 1990 to 2014 that can be characterized as nonviolent. Although the past few years may be indicating an increase in the occurrence of peaceful elections in Africa, the past 24 years of elec-

Electoral Violence in Sub-Saharan Africa 49

tions suggest that there is wide variation from year to year and decade to decade. In 2006, close to 71 percent of elections did not experience violence but just two years later, in 2008, only 25 percent of elections were non-violent. A meaningful trend has yet to emerge. We can be cautiously optimistic that the most recent years of peaceful elections will reproduce in the future, but that is far from certain. Since 1990, on average, 55 percent of elections have experienced some form of violence. Another way to examine trends in election violence is to examine if election violence decreases in specific countries over time. Perhaps election violence is a phenomenon that affects countries in the first few iterations of elections as actors learn new roles and adjust to new processes. Founding elections in particular may be prone to violence. Figure 2.7: Nonviolent Elections in Africa, 1990-2014

90   80   70   Percent

60   50   Non-Violent Elections

40   30   20   0  

1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

10  

There is some evidence to support this thesis. The majority of founding elections, or the first elections a country holds as part of a new dispensation, experience violence. According to the data, only 35 percent of the first elections held by countries in Africa have been peaceful. Each election a country holds seems to become slightly less likely to experience violence. By a country’s fifth election, slightly more than half (51 percent) are nonviolent. And of the countries in the sample that have held

50

The Political Context of Electoral Violence

  nine elections, only 27 percent of these elections exhibit some form of election violence. Of the sample of 47 cases, however, only 11 countries have held 9 successive elections. Some countries (Burkina Faso, Central African Republic, Guinea, Guinea Bissau, Madagascar, Mali, Mauritania, Niger) experienced coups d’état during this period while others have experienced significant conflict (Angola, Burundi, DRC, Liberia, Sierra Leone) that resulted in a change of regime. It is unclear how these conflict patterns and regime interruptions have affected the occurrence of electoral violence. Conversely, it is also unclear if and how electoral violence is related to the outbreak of larger conflict and regime change. Furthermore, democratic development and/or the expansion of civil and political liberties have stagnated in many countries. The relationship between election violence and democratic development is, at best, inconclusive. Looking at average Polity scores based on the number of elections a country has held, of the 11 countries that have now held nine elections since 1990, their average Polity score is 1.3. While this is double what the average Polity score was for most countries at the time of their first election (on average, 0.68), it is very similar to the average Polity score for countries in the sample at the time that their third election was held (on average, 1.3). In many of the country’s that have held nine elections, there was little change in the quality of democracy or the openness of electoral contestation between their third and ninth elections. Countries such as Cameroon, The Gambia, Madagascar, Togo and Zimbabwe—all of whom have held nine elections since 1990— have all shown little democratic growth over the past 24 years. Electoral violence may be preventing democracy from progressing, or it may be a symptom of a much deeper political malaise. Conclusions

Electoral violence has occurred in some form in roughly 55 percent of elections held in sub-Saharan Africa since 1990. It has the potential to significantly destabilize countries as has occurred in Côte d’Ivoire and Kenya; however, only 10 percent of elections in Africa take on such an extreme form of violence. Empirical work detailing the prevalence and nature of electoral violence is currently being undertaken but the causes of electoral violence are not yet well identified. Electoral violence is distinct from other forms of political violence in terms of its specific intent—namely, influencing electoral outcomes. As such, it can be considered one of many electoral strategies available to political aspirants in

Electoral Violence in Sub-Saharan Africa 51

the formulation of a political campaign meant to maximize the chances of winning an election. This chapter examined the factors that make electoral violence more or less attractive to politicians as a potential electoral strategy. The analysis suggests that electoral institutions can affect the incidence of electoral violence; that the electoral systems and the timing of elections matters; and that the political environment under which elections is extremely important in affecting the likelihood of violence taking place. The political environment also affects the type of election violence that most often occurs. Elections held under autocratic regimes are more violent and the violence is most often strategic. Elections held under transitional regimes experience slightly more incidental violence but strategic violence also takes place quite often. These findings point to two substantive conclusions regarding elections in less-democratic settings. First, peaceful elections require much more than the simple willingness of a regime to hold them. They require significant supporting architecture that is both autonomous and capable. The media must be able to observe and report on campaign activities. Elections must be observed and monitored by actors that have a broad base of trust from political elites and citizens alike. A respected dispute resolution mechanism must be firmly in place. There needs to be a holistic approach to election management that focuses on the entirety of the process, not just events on the day of elections. Second, elections are not, in and of themselves, an indicator of democracy. They are a necessary prerequisite but they are only one of many factors that contribute to the democratization process. The number of relatively durable electoral autocracies in Africa is a testament to this fact. Additionally, because democratic quality is a stronger and more consistent factor affecting incidences of electoral violence, this finding directs policy interventions in a potentially more fruitful direction. It may not be reasonable to suggest changing electoral rules as a remedy to electoral violence. Some countries, such as Nigeria and Kenya, have shown a willingness to moderately tweak existing institutions so as to prevent conflict around elections. Newly democratizing countries may also be willing to take into account the unintended consequences of various electoral regimes. Still, significant institutional change is a rare occurrence. Of the 48 countries examined in this chapter, only four have undertaken meaningful changes in electoral rules (Angola, Kenya, Lesotho, and Nigeria), three of which were a direct result of problematic elections. Focusing on the development of other supporting democratic institutions may be a more realistic approach to preventing and reducing levels of electoral violence in Africa.

52

The Political Context of Electoral Violence

  Notes

                                                                                                           

1 See for example Bekoe (ed), Voting in Fear; Collier, Wars, Guns and Votes; Daxecker, “The Cost of Exposing Cheating”; Dunning, “Fighting and Voting”; Fjelde and Höglund, “Electoral Institutions and Electoral Violence in sub-Saharan Africa”; Frazer and Gyimah-Boadi, Preventing Electoral Violence in Africa; Goldsmith, “Electoral Violence in Africa Revisited”; Höglund, “Electoral Violence in Conflict-Ridden Societies”; Obakhedo, “Curbing Electoral Violence in Nigeria”; and Straus, “It’s Sheer Horror Here.” 2 Straus and Taylor, “Democracy and Electoral Violence in Sub-Saharan Africa.” 3 Daxecker, “The Cost of Exposing Cheating,” 510. Data estimates on preand post-election conflict incidents and average number of conflict incidents come from replication data available at http://www.prio.org/jpr/datasets. 4 Goldsmith, “Electoral Violence in Africa Revisited,” 8. 5 Höglund. Kristine. 2009. “Electoral Violence in Conflict-Ridden Societies.” 6 Derçon and Gutierrez-Romero. “Triggers and Characteristics of the 2007 Kenyan Electoral Violence.” 7 Optiz. 8 Mueller, “The Political Economy of Kenya’s Crisis.” 9 Boone, “Politically Allocated Land Rights and the Geography of Electoral Violence,” and Rutten and Owuor, “Weapons of Mass Destruction.” 10 Obakhedo, “Curbing Electoral Violence in Nigeria,” 107. 11 Norris, “Why Electoral Malpractices Heighten Risks of Electoral Violence”; Weidmann et al., “Violence and Election Fraud.” 12 Kelley, Monitoring Democracy. 13 Straus and Taylor, “Democracy and Electoral Violence in Sub-Saharan Africa.” 14 See Chapter 3 for further discussion. 15 Straus and Taylor, “Democracy and Electoral Violence in Sub-Saharan Africa,” 29. 16 Bratton and Masunungure, “Zimbabwe’s Long Agony.” 17 Laakso,“Insights into Electoral Violence in Africa.” 18 “Kenya: the December 29, 1992 Elections.” 19 Kagwanja, “Facing Mount Kenya or Facing Mecca?” 20 Sisk, “Evaluating Election-Related Violence.” 21 Ogana, “Tanzania: Of CCM’s Economic Supremacy and Gangs.” 22 Aldrich, Why Parties?; Cox, Making Votes Count. 23 Duverger, Les Partis Politiques; Cox, Making Votes Count. 24 Powell, Elections as Instruments of Democracy. 25 Lijphart, “Constitutional Choices for New Democracies.” 26 Horowitz, A Democratic South Africa?; Lijphart,“Constitutional Choices for New Democracies.” 27 Cho and Bratton, “Electoral Institutions, Partisan Status, and Political Support in Lesotho.” 28 Mylonas and Roussias, “When Do Votes Count?” 29 Lindberg, Democracy and Elections in Africa.

Electoral Violence in Sub-Saharan Africa 53

 

30 Examples of violent harassment as defined by Straus and Taylor (2012) include “police or security forces breaking up rallies, party supporters fighting, street brawls, opposition newspapers being confiscated, candidate disqualifications, and limited short-term arrests of political opponents,” (p 10). 31 Straus and Taylor, “Democracy and Electoral Violence in Sub-Saharan Africa,” 10. 32 POLITY measures autocratic/democratic quality. The variable POLITY ranges from -10 (pure autocracy) to 10 (pure democracy) and is available for years, countries. 33 Freedom House scores range from 1 (Free) to 7 (Not Free) and are assigned based on expert surveys. 34 Mo Ibrahim ranges from 0 to 100, 0 being the worst score and 100 being the best. 35 Alabi, “The Legislatures in Africa.” 36 Prempeh, “Presidents Untamed.” 37 Bratton et. al., “Voting Intentions in Africa.”

3 Kenya: Strategic Electoral Violence

Due to the severity of the post-election crisis in 2007 and 2008, much has been written on Kenyan elections, but it is often overlooked that the 2007 election violence was not anomalous for the country. Violence has occurred in almost every election in Kenya—including the single party contests of the 1970s and 1980s. The perpetrators and motivations, however, have changed over time and in response to the environment in which elections have taken place. Elite-level political compacts, perceptions of electoral competitiveness and fraud, the quality and independence of ancillary institutions, and international intervention have all impacted the incidence and nature of electoral violence. The timing and severity have changed over time but there are two factors that have remained constant: much of the violence has been organized by political actors and rarely have perpetrators ever been prosecuted or punished. Violence has taken place in most, if not all, of Kenya’s post-colonial elections. In 1969, while Kenya was experimenting with multiparty elections in its immediate post-colonial phase, politician Tom Mboya was assassinated. Mboya was one of the founding members of the Kenya African National Union (KANU) and a minister in President Jomo Kenyatta’s government. He was a very charismatic figure and some speculated that held presidential aspirations. It has been alleged that his death was planned at the very highest levels of government in order to limit political competition ahead of the December 1969 elections.1 A few months after Mboya’s death, President Kenyatta attended a raucous political rally in Kisumu where security forces ended up shooting into the crowd, killing 10 and injuring 70.2 Shortly thereafter, Kenyatta announced that opposition parties were banned, effectively reducing Kenya to single party rule through the only remaining legal party, KANU.

55  

56

Kenya: Strategic Electoral Violence

In 1978 upon Kenyatta’s death, Vice President Daniel arap Moi assumed the presidency. According to Throup, elections became increasingly more violent in the 1980s.3 Moi employed henchmen to terrorize voters, in some cases dragging them to the polls to make sure that they voted the correct way.4 The transition back to multiparty elections in 1992, which Moi only agreed to under international pressure, did nothing to reduce electoral violence. Moi had argued that multiparty elections would encourage tribalism and lead to more political violence. In hindsight, this seems to have been a veiled threat: violence was perpetrated by Moi’s government before the 1992 and 1997 elections in order to intimidate and displace the opposition. Thousands died and hundreds of thousands were forcibly removed from their homes. In 2002, with Moi constitutionally barred from seeking a third term, violence occurred before the election but it was mostly sporadic and much less intense than previous contests. Before the 2007 election, violence took place but the most significant occurred after the election—much of it planned well in advance. Although widely praised for its “peaceful” character, the 2013 election can be linked to the deaths of at least 300 Kenyans in the period prior to the election. In comparison to 2007, it was a success. In comparison to a peaceful and well-managed election, the 2013 election failed on multiple accounts. This chapter provides an overview of Kenya’s political history, exploring the violent character of its electoral contests and the pivotal role ethnicity has played in Kenyan politics over time. Electoral violence has taken place in both single party and multiparty elections in the postcolonial period; the commonalities being the competitiveness of the elections, the use of single member districts (which encourage a winnertake-all mentality), and the impunity violence entrepreneurs enjoyed. The 2007 election—in which violence of multiple timings and types took place—is analyzed in more depth and from different perspectives in order to better identify the causes and triggers of electoral violence and the circumstances under which violence may escalate. Ethnicity and Politics in Kenya

Kenya was governed by some form of external British rule from the late 1800s until 1963. The decision to decolonize in the 1960s was preceded by a violent uprising the decade prior. During the 1950s, indigenous peoples fought against colonial rule in what was termed the “Mau Mau rebellion.” Many of the rebellion’s leadership were imprisoned, including Jomo Kenyatta, president of the Kenya African Union (KAU), a proindependence organization. The Mau Mau rebellion, in which upwards

Electoral Violence in Sub-Saharan Africa

57

of 25,000 Africans lost their lives, contributed to the British decision to grant independence to Kenya in the 1960s as part of a managed transition.5 The Kenya African National Union (KANU), formerly referred to as KAU, won a majority of parliamentary seats in pre-independence elections held in 1963. Upon independence, Kenyatta, president of KANU, assumed the position of prime minister of Kenya. Kenyatta began consolidating power almost immediately. Within a year, Kenyatta had the constitution amended so that Kenya became a presidential republic with expansive executive powers and Kenyatta its first president. Ethnicity has long been important for understanding politics in Kenya. At independence, it was estimated that Kenya had approximately 40 ethnic groups although more than half of the Kenyan population belonged to four groups: the Kikuyu, Luo, Luhya, and Kamba. Kenya’s largest ethnic group—the Kikuyu—has only ever comprised approximately 20 percent of the population, thus creating the necessity for some level of ethnic cooperation on the part of political elites. During the colonial period, the British governed the Kenyan territory through a policy referred to as “divide and rule” in which certain ethnic groups were privileged over others in order to keep conflict localized and to prevent unified opposition to colonial rule.6 The British reinforced ethnic differences and, in some cases, helped to create new ethnic groups. The Kalenjin and the Meru both combined several distinct tribes into single communities to better advocate for their collective preferences under colonialism.7 For the most part, divide and rule proved an effective method of administration as it prevented the Kenyan population from waging a sustained effort against colonial rule—that is until the Kikuyu-dominated Mau Mau rebellion erupted in the 1950s. Despite its importance in understanding Kenyan politics, ethnicity is hardly fixed; it is fluid and has changed based on the political imperatives of the times. For example, the Kalenjin was “created” in the 1940s as a merger of Nandi speakers and the Kipsigis in the Rift Valley as a counterbalance to the emerging threat the Kikuyu posed.8 During the 1950s, the British explicitly supported Kalenjin unity while the colonial government was fighting against the Mau Mau insurgency.9 The malleability of ethnic identity becomes even more apparent when looking at the ethnic alliances that have existed in multiple different permutations over time. In post-colonial Kenya, at times Luo leadership have allied with Kikuyu leadership against Kalenjin leadership; at other times Luo leadership have allied with Kalenjin leadership against the Kikuyu. These shifting alliances and coalitions contrast the view that Kenya’s ethnic feuds are hardened and long standing. Rather, they are strategic calculations that have changed depending on the political envi-

58

Kenya: Strategic Electoral Violence

ronment and elite-level preferences at a given time. Lending further credence to this interpretation of the nature of ethnicity, many (but not all) voters in Kenya have followed the lead of ethnic elites when casting their ballots.10 As Kenyan independence seemed more and more likely in the later 1950s, the British helped to plan and manage the transition. Prior to independence, the British assisted the Kenyans in creating administrative districts that were drawn around the existing ethnic settlements. For example, Nyanza Province comprised mostly Luo populations, Western Province was dominated by the Luhya, and Central Province was largely populated by Kikuyu. These ethnically-based provinces and districts have remained largely intact throughout post-colonial Kenya and have helped to promote ethnicity as a basis for political mobilization. While ethnic fragmentation is not inherently problematic, in Kenya it has been very important in explaining political pacts and elite-level alliances, in part, because ethnicity is an easy and obvious political identity to mobilize action around, especially when groups are geographically concentrated as is the case in Kenya. Even before independence was granted in 1963, two political movements loosely grouped along ethnic lines formed: the Kenya African National Union (KANU) and the Kenya African Democratic Union (KADU). Support for KANU came primarily from the Kikuyu and Luo while support for KADU was concentrated amongst several smaller ethnic groups in the Rift Valley Province, the Luhya of Western Province, and the Mijikenda of Coast Province. KANU was an alliance of the educated, urban, and more politically mobilized groups. The KADU base was mostly uneducated and rural. This fault line—of educated and wealthy Kikuyu (and their shifting allies) versus the smaller rural groups of the Rift Valley—persisted until at least 2007. KANU formed Kenya’s first independent government under the leadership of Jomo Kenyatta. As the first Kenyan president, Jomo Kenyatta, Kikuyu, consolidated power through the imposition of a de facto one party state. Kenyatta promised position and patronage to KADU members who defected, and in 1964 forced a merger between the two parties, eliminating KADU altogether. In 1966, former KADU leader Daniel arap Moi, Kalenjin, was appointed vice-president. Resentment of the merger led to the creation of the Kenyan People’s Union (KPU), which was headed by former vice president (1964-65) Jaramogi Oginga Odinga, prominent Luo and father to Raila Odinga. After several incidences of political violence broke out in 1969 including the assassination of prominent Luo politician Tom Mboya and a deadly political rally, the KPU was banned and its leaders were imprisoned.

Electoral Violence in Sub-Saharan Africa

59

During his time as president, Kenyatta played ethnic favoritism, almost exclusively appointing Kikuyu to important cabinet positions and other high-ranking offices such as the head of the central bank, civil service, and police. Kenyatta died in 1978 and was succeeded by Daniel arap Moi. Under Moi’s leadership, political and economic power shifted to the Kalenjin. Moi sought to overturn what he saw as an unequal distribution of wealth and resources in Kenya, and diverted public expenditure on roads, water, health, and education away from Kikuyudominated Central Province to the Kalenjin-dominated Rift Valley and other provinces. In 1982, Moi pushed through a constitutional amendment that made Kenya a de jure one party state. Despite these obvious political biases, the Kenyan economy fared relatively well under Kenyatta and Moi in the 1960s and 1970s.11 By the late 1980s, however, Moi’s mismanagement of the economy and his monopolization of power began to alienate large segments of the population. Senior members of the Kenyan political establishment and Moi’s inner circle called for political reform. The first was Kenneth Matiba, a Kikuyu cabinet member and businessman who resigned in 1988. In May 1990 Matiba and Charles Rubia, another minister, publically urged the return to multiparty politics. They were quickly detained, but riots broke out in Nairobi in July. Moi argued against multiparty elections, claiming that electoral competition would cause ethnic conflict and violence. In August 1991, Oginga Odinga, joined together with eight other opposition leaders including recently freed Matiba and Rubia, to form the Forum for the Restoration of Democracy (FORD). With the formation of FORD, the restoration of the early KANU Kikuyu-Luo alliance under Kenyatta returned. The international donor community also began to worry about the deterioration of politics and growing instability in Kenya, a donor success story until that point. The final push for the repeal on the constitutional ban on parties came in the form of a warning by Kenya’s bilateral donors at a November 1991 meeting of the Consultative Group (CG) in Paris. Convinced that economic growth required political reform, donors established political conditions for assistance and began directly funding Kenyan civil society. Two weeks after the CG meeting, Moi announced the ban on parties would be lifted immediately. And so began Kenya’s second period of multiparty elections. Elections, Ethnicity, and Violence in Kenya

During single party elections, violence was used primarily as a means of winnowing the field of competitors before elections. Several high-

60

Kenya: Strategic Electoral Violence

profile political assassinations and assassination attempts took place, and violence during the nominations process was common. Kenya has now held five elections since the resumption of multiparty politics in the early 1990s (1992, 1997, 2002, 2007, and 2013). Electoral violence has attended all five elections. To date, at least 4,000 Kenyans have died as a result of these elections and over 1 million have been displaced. There are three themes that reoccur throughout Kenya’s experience with elections and electoral violence: it is a strategy that has worked; political elites have manipulated ethnic “grievances” to encourage violence; and there have been few, if any, consequences for its use. In addition, and perhaps as a mechanism that reinforces the utility of the politicization of ethnicity in Kenyan election, political parties are fluid, partisan alliances are shifting, and political leaders seemingly call all the shots. Ethnic violence was reported across the country as soon as the 1992 campaign began. “Tribal clashes” erupted in Western Kenya as early as in December 1991; by May 1992, these clashes had spread to all areas along the borders of the Rift Valley Province, where non-Kalenjin migrants had settled adjacent to traditional Kalenjin lands. Violence was especially severe in the Rift Valley against people perceived as outsiders. Kalenjin “warrior” and Maasai “warrior” groups, dressed in traditional attire, attacked non-Kalenjin people in the Rift Valley, Nyanza, and Western Kenya, including Luo, Gusii, Luhya, Kamba, and Kikuyu.12 More than 1,500 people died and 300,000 were left homeless by the violence. The clashes were meant to show the mostly Kikuyu migrants that multiparty democracy would bring them trouble—that they could not just come into traditional KANU, or rather Kalenjin lands, and vote for the opposition. 13 As a political strategy, suppressive violence seems to have worked to prevent presumed opposition supporters from participating in the elections. The clashes succeeded in displacing many Kikuyu, disenfranchising them ahead of the polls and helping to elect Moi president with a simple plurality of the vote, 36 percent. Legislative elections were even more competitive as KANU won 100 legislative seats, more than 50 percent of those available, with only 24.5 percent of the total vote. All elections were conducted using simple plurality electoral rules in single member districts (i.e., the candidate with the most votes wins). As is often the case, violence was perpetrated in tandem with ballot box stuffing and various other attempts at vote rigging. The December 1997 election in many respects repeated the patterns of 1992; however, the opposition was even more divided. Fourteen candidates ran for president and 24 parties participated in the general election. The two FORD factions had disintegrated. FORD-Kenya, the 1992

Electoral Violence in Sub-Saharan Africa

61

coalition of the Luo and parts of the Luhya community, split when Luo political leader Raila Odinga formed the National Development Party (NDP). FORD-Kenya became a Luhya party led by Kijana Wamalwa. FORD-Asili also split when Kenneth Matiba refused to run for president. Five different Kikuyu ran for president but only Mwai Kibaki of the Democratic Party (DP) was considered a serious contender. President Moi won re-election with 40 percent and was the only candidate with more than 25 percent in five of the eight provinces. Kibaki received 30 percent of the vote and over 25 percent of votes in three provinces. The 1997 election took place with less violence and fewer deaths were reported, however, it was estimated that by March 1998, about 200 people had died in election related clashes. 14 Pre-election violence began in earnest in July 1997 when incidental clashes erupted between the state security apparatus and opposition groups.15 August and September 1997 saw serious ethnic clashes emerge on the Coast and in the southwest. These clashes were similar to the strategic clashes that had occurred in Rift Valley province in 1992, as violence sprees were carried out by large organized groups; were brutal, costing hundreds of people their lives; and targeted known opposition supporters.16 In August 1997, ‘Digo Warriors’ – financed by political candidates - at the Coast killed 100 and displaced 100,000 upcountry people.17 President Moi won the 1997 elections with 41 percent of the popular vote to Kibaki’s 31 percent. Moi’s party, KANU, performed better in legislative elections, winning slightly more than 50 percent in the 210seat National Assembly. In the Rift Valley, violence again targeted Kikuyu and non-native groups, with the Pokot, Samburu, and other ethnic minorities reportedly acting as aggressors.18 The justice system and the police forces appeared to lack impartiality—further threatening the Kikuyu populations in the Rift Valley.19 Many of the youths arrested for causing disturbances in the coastal area reported receiving training from and/or swearing an oath of loyalty to policemen and ex-security forces. Senior ruling provincial officials pardoned most of the KANU activists responsible for the unrest. 20 Additionally, more Kikuyus were tried for crimes committed against Kalenjin in the Rift Valley, even though more Kikuyus were arguably targeted. The 2002 Kenyan election presented a watershed moment in the nation’s history as incumbent Moi had been term-limited out of office and could not stand for president. On December 27, 2002, 5.8 million voters participated in the country’s election and ousted long-ruling KANU in favor of 15-party opposition coalition the National Alliance Rainbow Coalition (NARC). The NARC, led by veteran politician Kibaki, won 125 of 210 elected seats in the 224-member National Assembly. KANU

62

Kenya: Strategic Electoral Violence

representation plunged from 107 to 64 seats in the National Assembly.21 Kibaki won 62 percent of the vote and more than 35 percent in all eight provinces. His strongest support came from Nairobi, Western Province, and Eastern Province. Uhuru Kenyatta, the second runner up, performed best in KANU strongholds such as Rift Valley, where his image as Moi’s protégé and a guardian of Kalenjin interest made him popular. 22 Voter intimidation, bribery, and other irregularities were reported. Clashes in the border areas in western Kenya saw rural violence again. Odinga was roughed up by opponents while campaigning in Kisii and Uhuru Kenyatta’s sister reportedly handed out money to women in Gachoka constituency. But other than small clashes, intimidation, and bribery, the election was the most peaceful in modern Kenyan history.23 Many thought that Kenya had turned a corner and was heading toward democratic consolidation. After his election Kibaki appointed a 24person cabinet, much smaller than Moi’s 40-member cabinet. While Moi’s cabinet was composed of less educated and experienced members, Kibaki surrounded himself with skilled politicians. He did not, however, appoint Odinga to a cabinet position as promised for shoring up the NDP vote, creating a rift that would return with vengeance before the next election.24 After the relative successes of the 2002 election, the international community was shocked by the political crisis that unfolded in Kenya after the December 27, 2007 election. Large-scale violence erupted on December 30, 2007, three days after the election, when incumbent Kibaki was declared the winner. The violence appeared to take the form of ethnic conflict between those who had voted for Raila Odinga (Luo and Kalenjin) and those who voted for Kibaki (Kikuyu, Meru, and Embu). An estimated 1,500 people died and 600,000 were displaced, mainly in the Rift Valley, during the electoral violence, which raged until a power-sharing deal between Odinga and Kibaki was signed on February 28, 2008.25 Voting on Election Day, December 27, 2007, proceeded largely in an orderly fashion. As expected, the presidential race was very close. Polls conducted in the lead up to the election gave Odinga a slight advantage over Kibaki. Early returns put Odinga out ahead and confirmed that several members of Kibaki’s cabinet had lost their parliamentary seats. However, Odinga’s lead dwindled as results from Kibaki’s stronghold, Central Province, were announced on December 29. Then, unexpectedly, on the evening of December 30 the Electoral Commission of Kenya (ECK) announced a narrow victory for Kibaki, who was sworn in one hour later. The results that were announced gave 4.58 million votes to Kibaki versus 4.35 million for Odinga and 880,000 votes for

Electoral Violence in Sub-Saharan Africa

63

Musyoka of ODM-Kenya. 26 Two days later, ECK chairman Samuel Kivuitu told journalists he was no longer sure of who won the elections.27 Violence broke out almost immediately after the election and took several forms—rioting, incidental clashes between protesters and security forces, forced displacements, targeted killings, and reprisal attacks. First, after the ECK declared Kibaki the winner, seemingly spontaneous protests by ODM supporters erupted as they had expected a victory based on early results favoring Odinga. In urban Kenya, violence first started in Nairobi, Mombasa, Kisumu, and Nakuru. In rural Kenya, violence was concentrated in the Rift Valley, mostly in Eldoret. The initial wave of violence targeted those perceived to have voted for Kibaki, namely Kikuyus.28 Soon more organized attacks were noted to be taking place. PNU supporters started to organize counter-attacks, targeting mostly Luo. Kikuyu began to be violently evicted from Luo and Luhya territory. In the rural areas in particular, non-Kikuyus took the electoral violence that emerged in cities as an opportunity to chase the Kikuyus from their land. 29 Politicians took advantage of these sentiments to flush outsiders from their ancestral lands, which also conveniently served to eliminate the hostile vote. 30 In most cases, police were understaffed and ill equipped to handle the violence, or in the worst cases took sides and participated in the violence.31 During the violence, the government brought the paramilitary General Services Unit (GSU) into hotspots, which used excessive force, shooting over 100 people in Kisumu, in the middle of Odinga’s ethnic heartland. 32 Despite repeated calls for a cessation of hostilities, violence raged on throughout January and February. Estimates place the total number of dead during this period at close to 1,500. According to the Waki Commission, the international body created to investigate the 2007 - 2008 post-election violence, although some of the initial fatalities were the result of spontaneous rioting, many of the deaths were attributed to attacks planned by politicians and business leaders.33 This was especially true in Rift Valley, Central Province, and Nairobi.34 Since the jarring events of 2007/2008, there have been several explanations as to the underlying causes of Kenya’s election-related violence. Some have argued that the violence was an emotional response to highly competitive elections and the belief that the results were fraudulent.35 Others point to Kenya’s weak institutions, clientelistic parties, and centralized presidency as major contributing factors, creating an environment in which competition for the presidency was especially prized and raising doubts as to whether any political institution had the impartiality to fairly decide the outcome. 36 According to another expla-

64

Kenya: Strategic Electoral Violence

nation, the violence was fueled by rising expectations created by the increase in democratic space experienced during the 2002 election that pushed Kenyans over the edge when that space was curtailed in 2007.37 Still others believe that the violence was at its essence a land-based conflict.38 In order to understand the violence that occurred in 2007/2008, it is important to first describe its occurrence in more detail. First, the 2007 election was affected by both pre-election violence and post-election violence—the latter, of course, being much more intense and more deadly than the former. This is in part due to the differences in the goals of pre-election and post-election violence. Strategic pre-election violence is a coercive attempt to influence an electoral outcome before Election Day by influencing the behavior of candidates and voters as a part of the process. Strategic post-election violence is meant to change an electoral outcome after the fact through extra-judicial means, or those not previously agreed upon by political actors. Second, electoral violence did not occur evenly across the country. Thus, it is possible to analyze where electoral violence took place and if there are any environmental factors that might help to explain its occurrence. For the 2007 election Kenya’s eight provinces were subdivided into 68 districts. Each district contained between one and eight constituencies (matching legislative seats) for a total of 210 constituencies. There is significant variation in the size of electoral constituencies. Because delimitation of constituencies is not based on population, some constituencies had only 12,000 registered voters and some constituencies had almost 250,000 registered voters. Elections for the National Assembly took place at the constituency level. In order to examine the possible relationship between election violence and electoral competitiveness, Table 3.1 displays the frequency of pre-election and post-election violence across Kenya’s districts and provinces by the total vote support for the two main presidential candidates, Kibaki and Odinga. Data on electoral violence is taken from the Armed Conflict Location and Event Dataset.39 The dataset reports incidences of violence (including armed conflict between groups, destruction of property, riots, and protests) but does not describe the intensity or duration of events and does not include data on number of fatalities. Because there are several other types of events in the dataset (acts of terrorism, conflict over resources, rebellions, etc.) only events that took place in relation to the elections are included in the analysis.

Electoral Violence in Sub-Saharan Africa

65

Table 3.1: Electoral Violence and Vote Choice in Kenya’s Provinces, 2007 to 2008 Province

Majority Ethnic Group

PreElection Violence (Districts)

Central

Kikuyu

57%

Kibaki (97%), Odinga (1.5%)

14%

43%

Odinga (57%), Kibaki (35%)

43%

(7 districts)

Mijikenda, TaitaTaveta

Eastern

Somali

45%

Kibaki (52%), Odinga (13%)

0%

None

100%

Kibaki (48%), Odinga (44%)

100%

Somali

25%

Kibaki (52%), Odinga (47%)

0%

Luo and Kisii

42%

Odinga (79%), Kibaki (20%)

42%

Kalenjin

44%

Odinga (68%), Kibaki (30%)

50%

Luhya

50%

Odinga (71%), Kibaki (27%)

50%

(7districts)

Coast

(11 districts) Nairobi (1 district)

North Eastern (4 districts) Nyanza (12 districts) Rift Valley (18 districts) Western (8 districts)

Data taken from ACLED.

President Vote Percent

PostElection Violence (Districts)

66

Kenya: Strategic Electoral Violence

Although theoretically there are obvious differences between incidental violence and the various subsets of strategic violence (disruptive, suppressive, and mobilizing), it is often difficult to distinguish between these types without knowing the exact motivation of the actors. The data that is available reports the type of incident but rarely provides much insight into why exactly electoral violence takes place and what purpose it is meant to serve. Furthermore, because of the illicit nature of electoral violence and the clandestine methods political actors go about inciting violence, for the purposes of this analysis, electoral violence is only differentiated by timing and not intent. Nonetheless, looking at patterns of election-related violence and where exactly it is taking place can be revealing. Of Kenya’s 68 districts, slightly less than half (31) experienced some form of pre-electoral violence. Population size is positively related to incidences of electoral violence. Districts range in size from 62,000 inhabitants to close to 2.9 million residents. The larger more populous districts and constituencies more frequently experienced pre-election violence than did the less populated constituencies. Larger districts such as Kiambu, Mombasa, Nakuru, Bungoma, and Machakos more frequently experienced electoral violence than did smaller districts such as Lamu, Marsabit, Moyale, and Thakara. Of the 34 districts with a population size of less than 402,000 residents (the modal value), 24 percent experienced pre-election violence. Of the 34 districts with a population size of more than 402,000 residents, 68 percent experienced pre-election violence. This may be due to a number of factors including the number of aspirants vying for office increasing the competitiveness of the races, the amount of state resources available to legislators representing these districts, or the potential of affecting more voters in the aggregate. Pre-electoral violence occurred across Kenya’s eight provinces with no immediately obvious pattern. No province was spared the occurrence. At first glance, pre-electoral violence does not seem related to the competitiveness of the presidential election. Highly competitive provinces (Nairobi) and uncompetitive provinces (Central) were both venues for violence. It must be noted, however, that because pre-election violence is meant to affect the distribution of vote support, the relationship between competiveness and pre-election violence quickly becomes very difficult to disentangle. Districts and provinces may become more or less competitive as a direct result of pre-election violence. Because of this issue of endogeneity, it is important to look for information that is temporally distinct from the act of electoral violence and intimidation and also may be available to politicians. Politicians may rely on information from previous electoral contests in order to de-

Electoral Violence in Sub-Saharan Africa

67

termine where the most productive locations for violence and intimidation may be. Moving beyond the provincial level, at the district level (68 total districts) there is no statistically significant correlation between competiveness of the 2002 presidential election results and incidences of pre-election violence in 2007 (See Table 3.2).40 Table 3.2: Patterns of Competition and Pre-Electoral Violence in Kenya’s Districts, 2007 Presidential Election, 2002

Percentage Districts where Preelection Violence Took Place in 2007

0-10 Percent Margin of Victory

0%

(Highly Competitive)

(n=4)

0-25 Percent Margin of Victory

41.7% (n=24)

25-50 Percent Margin of Victory

55.5% (n=27)

50-75 Percent Margin of Victory

41.2% (n=17)

75-100 Percent Margin of Victory

33% (n=3)

Other possible factors affecting pre-electoral violence including poverty rate and support for Kibaki (in either 2002 or 2007) are also not significantly related to the violence. Interestingly, there is a small but significant relationship between competiveness in 2002 parliamentary elections and the incidence of electoral violence prior to the 2007 election. Electoral districts with more competitive legislative contests in the previous election were more prone to experience violence in the subsequent election in 2007. It seems likely that some of the pre-electoral violence in 2007 was a byproduct of legislative competitiveness rather than purely competition over the presidency. It should be noted that Kenya’s parliamentarians are amongst the highest paid in the Africa receiving approximately $US 10,000 untaxed per month in 2012.41 When the data is disaggregated even further to the constituency level (210 total constituencies examined), there is still no statistically significant relationship between pre-election violence in 2007 and the

68

Kenya: Strategic Electoral Violence

competitiveness of the previous presidential election, however, one potential explanation for this finding becomes much more apparent. Table 3.3 displays the percentage of constituencies that experienced preelection violence based on how competitive the previous presidential election was at the constituency level. Table 3.3: Patterns of Competition and Pre-Electoral Violence in Kenya’s Constituencies, 2007 Presidential Election, 2002

Percentage Constituencies where Pre-election Violence Took Place in 2007

0-10 Percent Margin of Victory

30.9%

(Highly Competitive)

(n=42)

0-25 Percent Margin of Victory

28.2% (n=85)

25-50 Percent Margin of Victory

30.2% (n=63)

50-75 Percent Margin of Victory

29.2% (n=48)

75-100 percent margin of victory

42.8% (n=21)

Of those constituencies considered highly competitive in the previous election, meaning that the margin of victory between the first and second top vote getters was less than 10 percent, 30.9 percent experienced pre-election violence in the following election in 2007. Of constituencies in which the previous election was highly non-competitive or dominated by one party or candidate over anther, almost 43 percent experienced some form of pre-election violence. In sum, the percentage of constituencies that experienced pre-election violence in 2007 remains relatively static regardless of how competitive the previous presidential election was. A partial explanation for this pattern has to do with the unmeasured and overlapping motivations for the use of pre-election violence. Specifically, suppressive electoral violence is generally targeted at suspected opposition voters and mobilizing electoral violence is generally targeted at one’s own supporters to increase vote support. Additionally, suppres-

Electoral Violence in Sub-Saharan Africa

69

sive and mobilizing election violence are both often used in the same election, even in the same region or district, and these types of strategic violence can produce a negligible effect in the aggregate, despite affecting individual level behavior and attitudes. This is a subject to which this book will return to in subsequent chapters. Incidences of post-election violence displayed some interesting patterns. Post-election violence did not affect Kenya evenly. Two provinces were spared from the violence that occurred after the election: Eastern and North Eastern. Both provinces voted marginally for Kibaki and both have majority Somali populations. Central Province, a Kibaki stronghold, experienced less post-election violence than most other provinces. While much of the violence (and the fatalities) was concentrated in Rift Valley Province, this was by no means the only place were post-election violence took place as Coast, Nairobi, Nyanza, and Western Provinces were also heavily affected by post-election violence. Support for Odinga (at the district level) was significantly related to post-election violence meaning that districts that supported Odinga recorded more violent events than did districts that did not support Odinga. This may be a reflection of the way that the violence after the election unraveled. Odinga partisans were protesting and rioting more frequently due to perceptions of fraud and retaliatory attacks after the protests and riots naturally targeted these areas. Pre-election and post-election violence are highly correlated at the district level: those districts that experienced pre-election violence were almost 50 percent more likely to experience post-election violence as those that experienced no such violence before the election.42 It could be that pre-election violence creates tensions that are easily primed again after an election. The implication for identifying areas susceptible to post-election violence would be to focus on those areas previously targeted for pre-electoral violence in the event that post-election violence breaks out. Table 3.4 reports data on fear of political violence across the 9 most populous ethnic groups in Kenya. The data come from an Afrobarometer survey conducted late in 2008 after the violence had subsided. The Afrobarometer survey is the largest cross-national poll of African attitudes regarding economic, political, and social issues. Its first survey was conducted between 1999 and 2001 in 12 countries (Botswana, Ghana, Lesotho, Malawi, Mali, Namibia, Nigeria, South Africa, Tanzania, Uganda, Zambia, and Zimbabwe). Since then, four additional surveys have been conducted, reaching more than 100,000 citizens in 35 different countries. Surveys have been conducted in Kenya in five waves or rounds thus far with additional surveys planned for the future.

70

Kenya: Strategic Electoral Violence

Table 3.4: Ethnicity and Fear of Political Violence in Kenya, 2008 Ethnic Group

Fear Political Violence “A Lot”

Kikuyu

37%

Luhya

40%

Kalenjin

26%

Luo

53%

Kamba

30%

Somali

35%

Kisii

56%

Mijikenda

31%

Meru

44%

All of the ethnic groups had a sizable portion of respondents reporting fear of political violence, but there are some striking differences across the groups. Of the 9 ethnic groups, Luo and Kisii respondents were most likely (53 percent and 56 percent, respectively) to report fearing political violence “a lot.” Of the Kikuyu surveyed, 37 percent reported fearing political violence a lot. Kalenjin respondents were almost half as likely (26 percent) as Luo respondents to report fearing political violence a lot. This gives support to the finding that while no one group was spared from the electoral violence that occurred in 2007-2008 in Kenya, Odinga’s Luo supporters may have been disproportionately affected or involved. Interestingly, members of other ethnicities aligned with Odinga (Luhya and Kalenjin, for example) did not report nearly as high levels of fear as did Luo respondents. Another way to contextualize the violence surrounding Kenya’s 2007 election is by examining the partisanship of the electorate. While it is easier to distill vote choice in Kenya as akin to an ethnic census, there is evidence that vote choice in 2007 was a combination of ethnic concerns and policy concerns.43 Approximately 50 percent of voters cast their ballot with their co-ethnics (in 2007 Kibaki received 94 percent of the Kikuyu vote, Odinga received 99 percent of the Luo vote, and Musyoka received 82 percent of the Kamba vote) but the remaining 50 percent of Kenyan voters used performance based criteria to inform their vote choice.44

Electoral Violence in Sub-Saharan Africa

71

Political parties in modern Kenya have typically been weak and fluid, as evidenced by the shifting party coalitions and factions of the 1990s and 2000s. NARC was no exception as it lasted only one electoral cycle. Part of its weakness can be attributed to its founding motivation: it was organized around what it opposed and not what it stood for. Kibaki’s absentee governing style compounded NARCs weakness. Moi micromanaged all aspects of governance, whereas Kibaki appointed people and gave them free reign to rule as they saw fit. The absence of a strong hand at the center of government meant that prominent leaders pursued their own agendas.45 Table 3.5 reports partisan identification (PID) over time across the 9 most populous ethnic groups in Kenya. It demonstrates how fluid and weak partisan identification is in Kenya and the shifting ethnic alliances that have taken place from 2002 to 2008. The slight majority of Kikuyu supported NARC but after the dissolution of the coalition and the events of 2007, only 38 percent of Kikuyu remained with Kibaki’s PNU. The Kalenjin, directly after the 2007 election, supported ODM in large numbers (73 percent, bold below) but a plurality had previously been part of NARC and KANU. The most striking feature of partisan identification in Kenya may be how few Kenyans actually report being close to any political party: only 50 percent of those surveyed identified with a political party. The weakness of partisan identification can most certainly be attributed to the weakness of political parties. Table 3.5 also depicts how political parties and coalitions in Kenya have changed quickly from election to election. NARC, the largest party/coalition in 2002, fell apart and led to the creation of a new political party, the PNU, ahead of the 2007 election. The PNU, with less than 2 percent political support in early 2013,46 entered into a political pact with The National Alliance (TNA). KANU, founded in the 1950s, and ruled Kenya from independence until 2002. Today it is virtually irrelevant. Because of the instability and lack of durability of political parties, voters are left searching for other criteria on which to base their vote resulting in weak partisan ties. The lack of continuity or durability of political parties means that parties are less important than their political leaders. Raila Odinga, for example, has been a member of KANU, Ford-Kenya, NDP, NARC, and ODM. Mwai Kibaki was a member of KANU, LDP, and PNU. Uhuru Kenyatta has been a member of KANU, PNU, and TNA.

72

Kenya: Strategic Electoral Violence

Table 3.5: PID and Ethnicity in Kenya, 2002 to 2008 Ethnic Group

PID (2002)

PID (2005)

PID (2008)

Kikuyu

NARC (56%); None (31%)

NARC (51%); None (34%)

PNU (38%); None (51%)

Luhya

NARC (40%); None (40%)

NARC (27%); None (46%)

ODM (53%); None (39%)

Kalenjin

NARC (46%); None (30%)

KANU (31%); None (39%)

ODM (73%); None (22%)

Luo

NARC (42%); None (32%)

LDP (52%); None (32%)

ODM (84%); None (14%)

Kamba

NARC (76%); None (17%)

NARC (42%); None (38%)

ODM-K (59%); None (23%)

Somali

NARC (42%); None (40%)

NARC (17%); None (58%)

ODM (51%); None (39%)

Kisii

NARC (38%); None (19%)

NARC (22%); None (46%)

ODM (54%); None (29%)

Mijikenda

NARC (46%); None (50%)

NARC (26%); None (48%)

ODM (47%); None (35%)

Meru

NARC (67%); None (28%)

NARC (80%); None (2%)

PNU (47%); None (24%)

Without a strong party apparatus to assist in voter mobilization, what do aspiring politicians use to mobilize voters? Pre-existing identities are especially attractive to politicians for several reasons. Social cleavages such as religion, race, language, and ethnicity are relatively easy to use as attachment to these groups is often deeply held and they may have pre-existing organizational structures that politicians can easily rely upon to drum up support. They are also generally estimable in size so that, if adequately mobilized, politicians can guess how much support they

Electoral Violence in Sub-Saharan Africa

73

can expect and calculate how much they believe they need. Ethnicity is an especially easy and obvious campaign tool when ethnic groups are as geographically concentrated as they are in Kenya. This relationship between weak parties, personalized elections, and the primacy of ethnicity is self-reinforcing. And it plays a role in understanding electoral violence in Kenya. But it alone does not explain Kenya’s long history of electoral violence. To what can we attribute the violence in 2007-2008? Because violence was not as concentrated as some anecdotal reports have suggested, it cannot be understood as simply the extension of a land dispute in the Rift Valley. Pre- and post-election violence affected approximately half of the country. While there is no doubt that this existing grievance/conflict could have been used to inflame tensions and mobilize individuals for violence, it is not the only or primary factor precipitating the violence. The competitiveness of legislative elections in the previous election mattered to some extent but the closeness of the presidential election seemingly did not, although between the alleged fraud that took place and the likely effects of pre-electoral violence to decrease the competitiveness of elections, it is impossible to rule this out as a trigger. Furthermore, electoral violence can be used to mobilize and demobilize voters making it very difficult to untangle how exactly violence relates to electoral competitiveness. It may seem intuitive that individuals who are completely confident in electoral victory would have no motivation to engage in strategic violence, however, aspirants may desire to increase their vote totals and margins of victory for other reasons. An overwhelming victory may be used as a signal to deter future challengers. It may also be a part of a larger, national campaign strategy to drum up support from less competitive regions in order to supplement lower vote totals in opposition strongholds. This may explain why at the constituency level electoral violence is related to competiveness of the legislative contests and not presidential contests. At the national level the 2007 presidential contest was extremely close with the two leading candidates polling within a razor thin margin of each other. Although the post-election violence of 2007-2008 was more serious and fatal than in previous elections, violence has accompanied all previous elections. In order to explain the persistence of electoral violence, pre and post, we need to look to factors that have also endured or persisted in Kenya. Plurality electoral rules and single member districts have been used for both single party and multiparty elections, but because this does not change over time it cannot account for variations in the level of violence. Additional likely culprits are weak political institutions—including political parties—but also the electoral commission

74

Kenya: Strategic Electoral Violence

(which has frequently been manipulated by the executive) and the relatively weak (in comparison to the executive) legislature and judiciary. Elections and Violence in Kenya, Post-2007

After almost two months of crisis and conflict, on February 28, 2008, Kibaki and Odinga signed a power-sharing agreement mediated by former UN Secretary General Kofi Annan, head of an African Union Panel of the Wise that also included former Tanzanian president Benjamin Mkapa and Mozambican politician Graça Machel. The National Accord established a coalition government in which cabinet ministers were appointed according to the strength of the parties in the National Assembly. Raila Odinga was appointed to the newly created post of prime minister. In an effort to promote peace and prevent future violence, both parties agreed that constitutional reform would take place before the next election. Constitutional reform had been promised several times over the past decade but it proved impossible for all vested parties to come to agreement. The 2007 - 2008 post-election violence made it clear that constitutional reform could no longer be ignored or postponed. In August 2010 the coalition government of Kenya presented a new constitution to Kenyans for approval. The draft that was submitted to the electorate was the product of almost 10 years of debate and deliberation. The process began in 2000 under Moi but was pushed aside during the final years of his presidency.47 When Kibaki and Odinga came together to form the National Rainbow Coalition (NARC) in order to contest the 2002 elections against KANU, both agreed that constitutional reform would be enacted within the first 100 days of Kibaki’s presidency. The new constitution would divide executive power between the president and a prime minister—a position that Odinga would presumably hold after the constitution’s approval. However, after NARC’s victory in the 2002 election, Kibaki and his political supporters walked back from their previous agreement with Odinga and his supporters. Over the next few years two different drafts were deliberated in parliament: the Bomas draft which Odinga supported, and its amended version, the Wako draft which Kibaki supported. The Wako draft was the draft that was ultimately approved by Parliament and put to referendum. In November 2005, Kenyans voted against the Wako draft constitution by a vote of 58 percent against, 42 percent in favor. Odinga and Uhuru Kenyatta, KANU’s defeated presidential candidate from the 2002 election, to campaign against the new constitution. Their coalition came to be known as the Orange Democratic Movement, so named because on

Electoral Violence in Sub-Saharan Africa

75

the ballot an orange represented a “no” vote and a banana represented a “yes” vote. Many believed this defeat to be a reflection of Kibaki’s waning popularity as he campaigned heavily in favor of the new constitution.48 Shortly after the vote, Kibaki dismissed his entire government. The political acrimony that followed set the stage for the devastating events of 2007 and 2008. Turnout for the 2010 referendum was robust as approximately 70 percent of registered voters participated. The constitutional referendum was a peaceful event, even in Rift Valley. William Ruto was one of the key figures pushing for rejection of the constitution but after its approval by 68 percent of voters, he urged his supporters to accept the new constitution.49 Major provisions include the reduction of power of the executive in favor of the legislature and judiciary, the decentralization of political power to increase the autonomy of local government, and the restructuring of election management through the creation of the Independent Elections and Boundaries Commission (IEBC) that replaced the ECK.50 The 2013 elections were held on March 4, 2013 under the new dispensation. The IEBC was tasked with a host of responsibilities including the registration of voters; the delimitation of electoral constituencies; the regulation of political parties; the settlement of electoral disputes; the registration of candidates; voter education; facilitation of the observation, monitoring, and evaluation of elections; regulation of candidate campaign spending; developing of a code of conduct for candidates and parties; and ensuring compliance with legislation on nomination of candidates by political parties. The registration exercise took place in November and December 2013. An estimated 14.3 million Kenyans registered to vote in this time period. Under the previous constitution, the country had been subdivided into 8 administrative provinces with a total of 210 electoral constituencies. The new constitution devolved power to newly created county-level governments. Provinces were eliminated and instead the country was divided into 47 counties, which encompassed a total of 290 constituencies. Because of the devolution of power, voters now cast ballots in 6 elections: president, national assembly, senate, women’s representative, county governor and county assembly. There was also a new requirement that the president-elect receive at least an absolute majority of votes (50% plus 1) and a minimum of 25 percent of the vote in at least half of the counties. It was hoped that these requirements would force presidential candidates to campaign for votes nationally as opposed to relying on specific ethnic groups for electoral support, something that was commonplace under the previous plurality rules.

76

Kenya: Strategic Electoral Violence

In early December, two key pre-election coalitions were announced. CORD, an alliance between ODM, Wiper Party, Ford-Kenya, and the Federal Party of Kenya, was formed with Raila Odinga as its presidential candidate and Wiper leader Kalonzo Musyoka as its deputy presidential candidate. The Jubilee Alliance brought together former political adversaries Deputy Prime Minister Uhuru Kenyatta of The National Alliance (TNA) and William Ruto of the United Republican Party (URP) along with the National Rainbow Coalition and the Republican Congress. Both Kenyatta and Ruto were facing trials at the International Criminal Court (ICC) in The Hague, The Netherlands, for their alleged roles in organizing the PEV. Unlike in 2007 where Kalenjin leadership and Luo leadership formed the basis of support for ODM, Kalenjin leadership allied with Kikuyu leadership prior to the 2013 election to form the Jubilee Alliance. Jubilee brought together Kenyatta, Kikuyu, and Ruto, Kalenjin in order to contest the election against a coalition of Luo, Kamba, and Luhya politicians in CORD. Odinga, a Luo, Musyoka, a Kamba, and Moses Wetangula, a Luhya, formed the core leadership of CORD. Many believed Jubilee to be a marriage of political convenience and an attempt to evade international prosecution, especially as Kenyatta and Ruto were accused of orchestrating violence against each other’s communities. The ICC trials became an important feature of the 2013 elections, and at one point they were even scheduled to begin at approximately the same time as the elections. When asked how the trial might affect his ability to hold office, Ruto stated, “I have no problem reporting to The Hague when required and running the affairs of Kenya…. This is an ICT [Information and Communications Technology] world. I can do many things on the Internet to ensure Kenya is running while I attend to my cases in The Hague.”51 According to a November 2012 survey, 41 percent of Kenyans polled believed that Kenyatta and Ruto should not contest the 2013 election due to their pending trial at the ICC; 39 percent believed they should contest but resign if they win and are found guilty; 20 percent believed that the ICC trial should be ignored completely.52 In early December Kofi Annan urged Kenyans not to vote for candidates facing trials at the ICC, provoking the ire of several politicians including Kenyatta.53 Pre-election polls suggested that the race was going to be close. At the end of February Ipsos Synovate, one of Kenya’s leading polling firms, released the results of a survey conducted in mid-February that found Odinga and Kenyatta in a statistical tie, receiving 44.4 percent and 44.8 percent of the vote, respectively.54 Ipsos Synovate concluded that

Electoral Violence in Sub-Saharan Africa

77

neither candidate had sufficient support to win outright in the first round and that a second round of voting was likely to take place. Despite the new constitution, the many reforms that took place and the various groups who were actively working to promote peace before the election, there were still fears that 2013 could mirror 2007 in terms of election violence. There were several incidences of pre-election violence in the months prior to the election. In Garissa and Eastleigh, a suburb of Nairobi, violent attacks took place against Kenyan security personnel and citizens alike.55 Although some have tried to tie these attacks to the elections, it is also possible that these attacks were part of the growing backlash against Kenyan military involvement in Somalia. In August, clashes in Tana River between Orma and Pokomo were responsible for over 100 fatalities. Many commentators believed that this conflict to be directly related to the parliamentary races in Garsen, Bura, and Galole constituencies.56 There is evidence that this violence was carefully orchestrated, that a fundraiser was held to finance it, and that a militia was specifically trained.57 In January, conflict again broke out in Tana River leaving more than 20 dead. Interestingly, no electoral violence, before or after the election, occurred in Tana River in 2007. In another incident Shem Kwega, a candidate for parliament from Kisumu and a political ally of Raila Odinga, was assassinated in late October 2012. In total, it is estimated that upwards of 300 died in the lead up to the election.58 On Election Day itself at least 13 were killed in disruptive violent events in Coast, reportedly perpetrated by a secessionist group intending to prevent voters from participating in the election.59 Around the rest of the country, however, violence was kept to a minimum. This was the case despite numerous logistical failures that resulted in a mass abandonment of the electronic balloting system for a manual tally. The process took several days to rectify, leaving voters and candidates confused and concerned. The delay in announcing the results was eerily similar to events preceding the 2007 election crisis. It wasn’t until March 9th that the Chairman of the IEBC Issack Hassan announced Uhuru Kenyatta winner of the 2013 presidential election. Kenyatta had won with 50.1 percent of the vote to Odinga’s 43.7 percent. While Kenyatta had a reasonably sized margin of victory over Odinga, because of the new electoral rules in which the winning candidate must receive 1 more than 50 percent of the total vote, in essence he won the election by a razor slim margin—some 8,000 votes allowed him to win outright in the first round instead of forcing a second round of voting.60 Unlike in 2007 when he told his supporters to take to the streets, this time Odinga urged his supporters to remain calm while he challenged

78

Kenya: Strategic Electoral Violence

the results through the courts system. The Supreme Court heard CORD’s petition, but quickly and unanimously upheld Kenyatta’s victory as fair and credible. Sporadic protests broke out and a few fatalities were recorded but Odinga’s acceptance of the decision and concession seemed to quash any significant protest movement.61 Although there have been substantial allegations since that he did not win an outright first round victory due to electoral irregularities and inflated vote totals, Kenyatta was sworn in as Kenya’s fourth post-independence president on April 9, 2013.62 Why did extreme electoral violence not manifest in the 2013 election? First, it is important to contextualize 2013 alongside Kenya’s previous electoral contests. Pre-electoral violence of varying degrees has occurred before every multiparty election to date. Although there were some incidences of post-election violence after the 1992 election and the 2013 election, post-election violence of the magnitude witnessed after the 2007 election was unique to that election. There are some commonalities between 2007 and 2013 however; namely, weak political parties, weak (although new) institutions, a delay in announcing the results, allegations of electoral impropriety and a very close, competitive election. Significant pre-election violence took place before both the 2007 and 2013 elections. There are major differences between 2007 and 2013, however, which helped to prevent a reoccurrence of large-scale postelection violence. The Kikuyu/Kalenjin political coalition seemed to mitigate any remaining ethnic antagonisms left over from 2007. Additionally, the fact that no true incumbent ran for president (as was the case in 2002) means that it would be more difficult to manipulate state resources in favor of one candidate over another as it is believed that Kibaki was able to do in 2007. The increased attention focused on Kenya’s elections also made large-scale post-election violence as was seen in 2007 and 2008 difficult to organize. This was a high profile election with many international election observers from the US, the European Union, the African Union and the East African Community, among others, dispatched to various regions of Kenya. International and domestic media scrutinized every action and speech delivered by political leaders. Websites such as Uchaguzi, Frontline SMS, and Ma Vulture were created to use crowd sourcing to monitor electoral violence. It would have been extremely difficult to plan and prepare for widespread violence without detection. In 2007, this was not the case. Because of the relative success of the 2002 election in terms of peacefulness and the turnover of political power from the long-ruling KANU to NARC, many believed that Kenya had matured and its elections were no longer so polarizing and acrimonious.

Electoral Violence in Sub-Saharan Africa

79

Many assumed the 2007 election would be a rote exercise in democracy. In addition, according to Kelley (2012) international election monitoring was inconsistent during elections held between 1992 and 2007.63 No organization monitored more than two polls during this period. It was not until the events of the 2007 and 2008 that organizations realized how dangerous and deadly elections in Kenya could become. The international community was exceptionally active in working to prevent electoral violence from taking place in 2013. The United States government alone spent more than $150 million on democracy and governance programs in Kenya in the five-year period prior to the elections.64 The International Foundation for Electoral Systems (IFES) worked in concert with the United States Agency for International Development (USAID) and the IEBC on election preparations. There were countless international non-governmental organizations working across the country to prepare for the elections including the Carter Center, Human Rights Watch, International Crisis Group, and the Electoral Institute for Sustainable Democracy in Africa (EISA), to name just a few. Domestic civil society was also heavily involved in election preparation. The Elections Observation Group (ELOG), a consortium of civil society organizations and other stakeholders, was established in 2010 to “provide a permanent national platform though which citizens can monitor general elections in Kenya.”65 The body’s first project was the monitoring of the 2010 constitutional referendum in which they deployed over 10,000 observers. For the 2013 elections, ELOG observed and monitored the periods prior to, during, and following the elections. According to the group, there were significant efforts (including a simulation exercise) to educate political parties, the IEBC, and other key stakeholders about what expect on Election Day. PeaceNet is a national umbrella organization of non-government organizations (NGOs), community-based organizations, donor agencies, and individuals committed to conflict resolution and peace-building in Kenya. In cooperation with other partners, PeaceNet engaged in Peace Initiative Kenya (PIK) – a 15-month project concentrating on violence prevention in Kenya’s conflict zones during the pre- and post-election period.66 The Electoral Institute for Sustainable Democracy in Africa (EISA) organized workshops around the country to discuss early warnings of and early responses to potential electoral violence. The organization also assembled regional conflict management panels to assist in designing programs to prevent electoral violence based conflict drivers specific to each area. These are but a few examples of various programs undertaken in Kenya prior to the 2013 elections so as to prevent a repeat of 2007.

80

Kenya: Strategic Electoral Violence

Finally, the Kenyan government also established several organizations to monitor and respond to any indication of election-related violence. The Kenyan National Commission on Human Rights (KNCHR) is an autonomous institution established by Article 59 of Kenya’s 2010 constitution that acts as a watchdog over the government.67 Following the post-election violence of 2007-2008, the KNCHRs predecessor body (of the same name) released a report, “On the Brink of the Precipice: a Human Rights Account of Kenya’s Post-2007 Election Violence.” In it, the commission examined the causes and motivations for the violence. They concluded that significant planning and coordination took place during the attacks, many of which seemed to have been “carried out with military discipline.”68 The National Cohesion and Integration Commission (NCIC) is a statutory body established under the National Cohesion and Integration Act of 2008 directed to promote harmony, peaceful coexistence, and national integration. Prior to the election, NCIC engaged in civic education across the country and extensive hate speech monitoring. NCIC was also involved in election monitoring and had representative as various polling stations and the national tallying center during the primaries and general elections. Conclusions

The conduct of the 2013 elections showed a marked improvement from that of the 2007 elections and compared to 2007, the 2013 elections appear “successful.” However, it remains to be seen how durable these gains will be. Public opinion leading up to the election was divided between those who were optimistic that the elections would be free and fair and those who were more skeptical. Much of this optimism can be attributed to the reforms that had taken place since 2008: specifically, the newly established IEBC, the major reorganization of the judiciary, and the hoped-for effects of devolution. Unfortunately, while the conduct of the election was more peaceful than prior contests, in terms of electoral administration, there were still several irregularities that took place as part of the electoral process. Whether through incompetence or malfeasance, there were failures in electoral management on multiple fronts, especially when it came to the use of technology in the election.69 In a country where much of the rural population is off the electrical grid and the areas that do have access to electricity routinely experience blackouts, there was no contingency plan in case electronic devices failed or could not be accessed on Election Day. Several polling stations, it turned out, did not have electricity. There were also battery issues with many of the voting devices. Accord-

Electoral Violence in Sub-Saharan Africa

81

ing to ELOG, 93.5 percent of the polling stations had challenges using the poll books.70 The results transmission system also experienced problems during tallying. The IEBC procured 32,000 specially configured cell phone devices to electronically send provisional vote results from polling stations to a dedicated IEBC server. In many cases, the phones did not work due to battery issues. In addition, “low disk space” on the IEBC server led to a delay in the transmission of results from polling stations countrywide. Officials were forced to transport ballots to the national tallying center and count votes by hand, causing significant delays in the announcement of results. These mistakes did not go unnoticed by the Kenyan population and much of the blame can be and has been attributed to the IEBC who, as of almost a year after the election, had yet to release the official final vote totals at the polling place level. According to a February 2013 Ipsos Synovate poll 89 percent of those surveyed had some or a lot of confidence in the IEBC.71 As of February 2014, this number had decreased substantially with only 21 percent of respondents stating they had “a lot of confidence” in the IEBC and 25 percent stating that they had no confidence in the IEBC.72 Compounding the Commission’s issues, in late October, four officials from the IEBC, including IEBC Chief Executive Officer James Oswago, were arrested on corruption charges in relation to the procurement of the biometric voting kits used in the election.73 These developments do not bode well for future elections. Violence, while a feature of the 2013 election, did not escalate to previous levels for a number of reasons including increased scrutiny over the conduct of the election, a concerted effort by multiple actors to prevent violence from taking place, and a belief by most political actors that the process would be fair and transparent. If any of these factors are absent in the next elections, violence before and after is an almost certainty. Kenyan elections have, historically, been violent affairs. Despite the many accolades and praise the 2013 elections received, it was still a violent election. Strategic violence has taken place before, during and after most multiparty elections. While it is believed that previous presidents have organized violence, there is also evidence that political parties, parliamentarians and local politicians have also used violence as an electoral strategy. In addition to anecdotal evidence, the pre-election violence that took place prior to the 2007 cannot be directly linked to the competitiveness of the previous presidential election but it can be connected to the competitiveness of previous parliamentary elections. There are different interpretations to this finding, but it could be tied to the fact that violence in Kenya is not just a tactic used by incumbent governments but by a myriad of political actors.

82

Kenya: Strategic Electoral Violence

In terms of understanding the costs and benefits of the use of electoral violence in the Kenyan case, the consequences seem to be minimal. Very few, if any, of the alleged perpetrators and organizers of election violence have been prosecuted. There was such little political will in the country to bring some of the key parties responsible for the PEV to trial that the National Assembly voted down the establishment of a local tribunal.74 The cases were ultimately turned over to the ICC because the Kenyan government refused to prosecute any of the accused. When the costs of using violence are this low, the decision to strategically employ election violence or not can be influenced by factors such as individually held beliefs about the justness of violence and one’s own commitment to peace. It is not clear that either of these is universal. Notes

                                                                                                            1

“Foreign Agents Link in Mboya Assassination.” Hyden and Leys, “Elections and Politics in Single-Party Systems.” 3 Throup, David. 1993. “Elections and Political Legitimacy in Kenya.” 4 Throup, “Elections and Political Legitimacy in Kenya,” p. 384. 5 “Mau Mau Uprising.” 6 Omolo, “Political Ethnicity in the Democratisation Process in Kenya.” 7 Lynch, I Say to You. 8 Lynch, I Say to You. 9 Lynch, I Say to You. 10 Oyugi, “Ethnicity in the Electoral Process”; Bratton and Kimenyi, “Voting in Kenya”; Wanyama et al., “Ethncity and/or Issues?” 11 Mosely, Paul. “Kenya in the 1970s: a Review Article.” 12 Kagwanja, “Facing Mount Kenya or Facing Mecca?” 13 Barkan, “Kenya: Lessons from a Flawed Election.”   14 “Kenya Post-Election Political Violence.” 15 Ajulu, “Kenya’s Democratic Experiment: The 1997 Elections,” p. 276. 16 Foeken and Dietz, “Of Ethnicity Manipulation and Observation,” p.142. 17 Kagwanja, “Facing Mount Kenya or Facing Mecca?”   18 “Kenya Post-Election Political Violence.” 19 “Kenya Post Election Political Violence.” 20 Ajulu, “Kenya’s Democratic Experiment,” p. 275. 21 Ndegwa, “Kenya: Third Time Lucky?” 22 Anderson, “Briefing: Kenya’s Elections 2002.” 23 Anderson, “Briefing: Kenya’s Elections 2002.” 24 Anderson, “Briefing: Kenya’s Elections 2002.” 25 Haeneit-Sievers and Peters, “Kenya’s 2007 General Election and its Aftershock.” 26 Haeneit-Sievers and Peters, “Kenya’s 2007 General Election and its Aftershock.”   2

Electoral Violence in Sub-Saharan Africa

 

83

27 Haeneit-Sievers and Peters, “Kenya’s 2007 General Election and its Aftershock.” 28 Rutten and Owuor. “Weapons of Mass Destruction.” 29 Rutten and Owuor. “Weapons of Mass Destruction,” p. 319. 30 Rutten and Owuor. “Weapons of Mass Destruction.” 31 Rutten and Owuor. “Weapons of Mass Destruction.” 32 Haeneit-Sievers and Peters, “Kenya’s 2007 General Election and its Aftershock.” 33 The Waki Report. 34 The Waki Report. 35 Derçon and Gutierrez-Romero. “Triggers and Characteristics of the 2007 Kenyan Electoral Violence.” 36 Mueller, “The Political Economy of Kenya’s Crisis.” 37 Rutten and Owuor. “Weapons of Mass Destruction.” 38 Boone, “Politically Allocated Land Rights and the Geography of Electoral Violence.” 39 Raleigh et al., ACLED. 40 Based on results derived from logistic regression analysis. 41  “Kenya President Mwai Kibaki Rejects MPs’ Bonus Attempt.”   42 Based on results derived from logistic regression analysis. 43 Bratton and Kimenyi. “Voting in Kenya.” 44 Long et al., “Choosing Peace over Democracy.” 45 Barkan, Joel D. 2004. “Kenya after Moi.”   46 “Presidential Alliances February 2013,” p. 33.   47 Kramon and Posner, “Kenya’s New Constitution.” 48 “Kenyans Reject New Constitution.” 49 Kramon and Posner, “Kenya’s New Constitution.” 50 Kramon and Posner, “Kenya’s New Constitution.” 51 “I Can Run the Country Using ICT while at The Hague, Says Ruto.” 52 “SPEC Barometer Survey, November 2014.” 53 Mosoku, “Leaders Prop Mudavadi as Compromise Candidate.” 54 “None of the Candidate’s Garners 50%+1 for Clear Win.” 55 Ombati, “Fresh Violence Rocks Garissa”; “Blasts Rock Nairobi Neighborhood.” 56 Babo, “Tana River Violence Linked to Next Elections.” 57 Ndonga, “Kenya: Militia Was Trained for Tana Killings.” 58 “High Stakes: Political Violence and the 2013 Elections in Kenya” 59 “Kenyan Elections Marred by Mombasa Violence.” 60 Gettleman, “Voting Irregularities in Kenya Election Are Confirmed.” 61 “Kenya Court Upholds Uhuru Kenyatta’s Poll Win.” 62 Ferree et al., “Voting Behavior and Electoral Irregularities in Kenya’s 2013 Election.” 63 Kelley, Monitoring Democracy. 64 “USAID Support for Kenya’s 2013 Elections.” 65 “History,” Elections Observation Group; Personal communication, September 2013. 66 “Peace Initiative Kenya.” 67 Personal communication, September 2013. 68 “On the Brink of the Precipice,” p. 7.

84

 

Kenya: Strategic Electoral Violence

69 Bowman and Longwe. “Technology, Transparency, and the Kenyan General Election of 2013.” 70 Personal communication, September 2013. 71 “Presidential Alliances February 2013,” p. 21. 72 “ SPEC Barometer Survey, March 2014,” p. 15. 73 Ombati, “IEBC Top Officials Arrested over Sh1.3b Poll Kits Tender.” 74 “International Justice Monitor.”

4 Senegal: Incidental Electoral Violence

Senegal, located in West Africa and home to almost 14 million residents, enjoys a reputation as one of the most peaceful and stable countries on the continent. Senegal’s reputation for peace and tranquility is, however, based on events that have not taken place in the country. Although there was an alleged attempt in 1962, it is one of the only countries in sub-Saharan Africa to have not experienced a successful coup. It is ethnically and religiously homogenous as approximately 40 percent of the country is Wolof and almost than 90 percent of Senegalese is Muslim. Unlike its neighbors, Senegal has also never experienced a civil war, despite a low-grade insurgency in the Casamance region in southern Senegal that has been simmering since the 1980s. Additionally, it also has the unique ability to boast of three peaceful transfers of power in executive leadership, with two of these changes taking place between different political parties. It is against this backdrop, that the election violence of 2012 took the world by surprise. Commentators decried the “atypical”1 and “unexpected”2 violence that occurred in January 2012 ahead of the February presidential election. Many were concerned that mounting domestic unrest would cause significant political instability and derail democratic progress in a country viewed as one of Africa’s success stories.3 Some were worried about the political climate in Senegal and worried it could go the way of some of its continental counterparts, such as Kenya and Zimbabwe. Concerns were at a fever pitch, and then, after the concession of defeat by the incumbent president in March, the narrative quickly shifted back to praise for Senegal’s ability to peacefully resolve its political conflict. It is now touted as an example of the triumph of democracy in Africa and its reputation for peace and stability has been restored.4 It is helpful to remember that after the 2000 election many 85  

86

Senegal: Incidental Electoral Violence

praised Senegal for consolidating its democracy and showing democratic maturity.5 The opposition victory in 2000 and the peaceful turnover of political power belied the weaknesses of its institutions and the autocratic tendencies of its leader, Abdoulaye Wade. Elections in Senegal have historically been characterized by manipulation, fraud, and in some cases violence. The 2012 elections were arguably the most violent in Senegalese history with at least 6 dead and hundreds injured during the month leading up to the first round of elections held February 27, 2012. Unlike electoral violence in Kenya, however, the violence in Senegal was not organized by political actors as part of larger campaign strategy, but rather borne out of frustration and happenstance. The violence surrounding the 2012 election can be characterized as “incidental” as much of it occurred as a result of protesters engaging state security forces in disputes over the fairness of the electoral process. Because of the non-strategic nature of the violence in Senegal, electoral institutions do not play the same role in incentivizing violence that they do in cases of strategic violence such as found in Kenya, but fairness of the overall electoral process does play a role. One of the major triggers of the 2012 Senegalese electoral violence was the attempted power grab by an unpopular incumbent politician, Abdoulaye Wade, and the belief that he would go to almost any end to secure entrenchment. The violence eventually defused and de-escalated, but only after the first round election results were respected by Wade, signaling to the opposition that further attempts to remain in power were unlikely. This chapter addresses the following questions: why did violence take place, could it have been predicted, and why did it not escalate? In order to do this, I trace the history of elections and violence in Senegal. From this analysis some striking patterns emerge. First, electoral violence in several different forms has taken place in many elections in Senegal’s history. Second, violence and intimidation in the Senegalese case differs in some key respects from the more high-profile cases such as Nigeria, Kenya, and Zimbabwe, but it is a persistent trend nonetheless. Finally, there has been a pattern of electoral fraud, violence and reform that has repeated in at least 5 different iterations since 1960. In some instances, the reforms have resulted in a contraction of political freedom and sometimes the reforms have resulted in an expansion of political freedom. Identifying and understanding the pressures that ultimately lead to either contractions or expansions in freedoms provides insight into the incremental process of democratization and offers lessons for other newly democratizing countries. The Senegalese elections underscore both lessons for the resolution of electoral conflict and also

Electoral Violence in Sub-Saharan Africa

87

how the receding of democratic practices can create the necessary conditions for election-related violence to take place. History of Senegalese Elections

The Senegalese experience with elections dates back to the mid-1800s while it was still a French colony.6 Elections were held at irregular intervals and the franchise was limited, based on property ownership and citizenship criteria that evolved over time.7 Throughout the colonial period, elections were held in several larger cities including Dakar and Saint Louis for territorial deputies to represent colonial interests to the French legislature. In the late 1880s, the colonial administration organized elections for mayor and local representative bodies in several towns across the Senegalese territory, allowing African residents limited participation. In 1895, France united its eight colonies in West Africa, including Senegal, into a federation termed French West Africa. The capitol was located in Dakar, the present-day capitol of Senegal. A small indigenous political class emerged, many of who were educated in France and held various positions in the colonial bureaucracy. Regime instability and post-war reconstruction in France directly affected the manner in which it governed its colonies. France made a series of concessions meant to stave off full independence. In 1946, a new French constitution established territorial councils in all of its colonial possessions. French West Africa was incorporated into the French Union. A Senegalese territorial council was established, comprised of 50 elected seats. One party, the Bloc Union Republicaine Socialiste, won all 50 seats. As the French debated the future of colonialism and the potential for independence amongst its overseas possessions, nascent multiparty politics flourished briefly in the 1950s in the Senegalese territory. Elections were closely contested and sporadic violence during the electoral process took place.8 After the promulgation of another new French constitution in 1958, the French Union became the French Community and each territory formally became a protectorate of France with slightly more autonomy including an elected legislative assembly. Each territory held a constitutional referendum to approve the new arrangement. A vote in favor of the new constitution meant continued association with France while a vote against translated into immediate independence. Of its African colonies, only Guinea voted against the constitutional referendum, effectively ending all ties with France.

88

Senegal: Incidental Electoral Violence

Although in Senegal the constitution was approved with approximately 98 percent voting in favor, a schism between Senegalese elites emerged between those that wanted immediate independence from France and those that wanted continued association. This divide heavily influenced political party formation and electoral competition in the late 1950s. Léopold Sédar Senghor formed the pro-France Senegalese Progressive Union (UPS) prior to assembly elections in 1959. Two other political parties, the Parti Africain de l’Independence (PAI) and the Parti du Regroupement Africain-Senegal (PRA), contested the elections and agitated for the “Africanization” of the territory.9 The UPS won all 80 Assembly seats. As the anti-colonization tide swept over Africa and independence appeared imminent, Senghor opted for a managed transition to independence. Senegal experienced a peaceful decolonization process with power handed over from the French colonial administration to its first post-independence government under Senghor. Even after independence, Senghor opted to maintain close economic and political ties with France.10 Senghor was a pragmatic politician with strong allegiances to France. Born in the city of Joal south of Dakar, he was a poet and linguist. He had spent close to 20 years working and living in France. In 1935, he graduated from the University of Paris and took a position as professor in Tours, a city in central France. He also served in the French army during World War II where he was captured by German troops and held as prisoner of war for approximately 18 months. He was released in 1942. He began his Senegalese political career in 1946 when he was elected to the territorial council as part of the Bloc Union Republicaine Socialiste. Since independence, Senegal has held regular elections and, despite one alleged coup attempt in 1962, has never experienced adverse regime change. Senghor, an elected politician under the colonial regime, was also elected Senegal’s first president and remained so until his voluntary retirement in 1980, another rarity in Africa at the time. During his two decades as president, Senghor allowed for elections but imposed various restrictions on party competition. Unsurprisingly, these elections varied in conduct and quality. When Senegal was granted independence in 1960, the country initially operated as a multiparty parliamentary system with Senghor elected as president and Mamadou Dia his prime minister. Violent riots in Saint Louis attended the 1960 elections, however, and after which the leaders of the PAI were arrested and charged with incitement.11 In December 1962, after a schism between the former allies, Senghor accused Dia of attempting a constitutional coup. Dia and several of his support-

Electoral Violence in Sub-Saharan Africa

89

ers were charged with treason, convicted and sentenced to life imprisonment. Dia ultimately was pardoned by the state in 1974. To address this political crisis, Senegal adopted a new constitution and established itself as a presidential system. Senghor was then elected president in 1963’s general election with 100 percent of the vote.12 His party, UPS, won all 80 seats in the National Assembly with 94.2 percent of the popular vote.13 Student riots before the election resulted in 50 or so students being shot by security forces and at least 10 deaths. The government arrested several political opposition leaders, with some choosing exile over imprisonment.14 Over the next few years, the political space closed considerably. Other political parties including the PRA were either co-opted and chose to join UPS or suppressed and “chose” to disband. Elections held in 1968 and 1973 saw Senghor and UPS run unopposed. During this period turnout is reported to be very high, averaging 80 to 90 percent, but both elections were reportedly “marked by violent incidents.”15 In the mid-1970s, riots and protests broke out in several rural areas over the country’s economic performance. 16 As a method to address the political protests and demands for reform, prior to presidential and legislative elections in 1978 Senghor introduced a novel piece of legislation that mandated the expansion of the Senegalese party system to include three ideologically designated parties. The Socialist Party (PS), successor to Senghor’s UPS, would represent the “Socialist” position, the Senegalese Democratic Party (PDS) would represent the “Liberal” position, and the African Independence Party (PAI) would represent the “MarxistLeninist” position.17 Another relatively popular party, the Rassemblement National Democratique (RND), was prevented from participating in elections, as it did not fit within the established ideological space. In the 1978 election, Senghor was elected president with 82 percent of the vote. The opposition candidate, Abdoulaye Wade from PDS, received the remaining 18 percent of the vote. The 1978 election marked the first of Wade’s four unsuccessful bids for the Senegalese presidency before succeeding in 2000. The PDS won 18 seats in the National Assembly (out of 100) but, alleging electoral fraud, refused to accept them.18 Compared to previous contests, turnout was a relatively low 63 percent, which RND leadership argued was due to the abstention of their supporters as they had called for an boycott of the election.19 Alleging fraud, the opposition challenged the outcome in court but the results were ultimately upheld. Under Senghor’s leadership Senegal held 5 general elections—in 1960, 1963, 1968, 1973, and 1978. Each election had reports of riots and protests, with mass arrests and exiles taking place after the 1960 election

90

Senegal: Incidental Electoral Violence

and multiple fatalities occurring during the 1963 election. After the violence that took place during the 1960 and 1963 elections, the political space contracted sharply and by 1968, Senegal had become a de facto one party state. Protests and calls for reform strengthened after the 1973 election in which Senghor and the UPS were returned to office with 100 percent support and an astronomical 97 percent turnout. Amid a faltering economy, Senghor and the PS (the UPS changed its name to the Parti Socialiste in 1976) made several concessions and reintroduced multiparty competition ahead of the 1978 elections, but with strict limitations on party formation and party content. On the last day of 1980, Senghor stepped down and turned over the presidency to his prime minister, Abdou Diouf. Diouf lifted the remaining restrictions on party formation and operation; however, it has been remarked that he merely tolerated the opposition and that under his presidency there was no true electoral competition as he imposed different yet more restrictive regulations.20 Nonetheless, political parties proliferated in the 1980s. Eight political parties participated in the 1983 elections, although only three were awarded assembly seats. There were no significant reports of violence as a part of the election but fraud was alleged and the results were again challenged in court, to no avail. Six parties participated in the 1988 elections but only two parties (PS and PDS) received sufficient vote support to earn seats. The 1988 elections were especially problematic as riots and protests broke out after the results were announced and Wade and the opposition claimed fraud. Wade was charged with inciting violence and brought to trial. Protests continued throughout Wade’s trial until President Diouf called for amnesty and agreed to engage in discussions with the opposition on issues including electoral reform.21 While the initial talks stalled, after the opposition boycotted local elections in 1990—and violent protests followed these elections as well22—Diouf invited Wade and other prominent members of the opposition to form a government of national unity.23 He also impaneled a commission with members drawn from all political parties to reopen discussions on electoral reform. These discussions resulted in sweeping reforms, many of which were meant to eliminate opportunities for fraud.24 Reforms included the adoption of the secret ballot, the use of indelible ink to prevent voters from casting more than one ballot, and the incorporation of representatives from opposition political parties at all stages of the electoral process.25 It was hoped that these reforms would result in more transparent and better managed elections but, unfortunately, the 1993 elections were marked by angry mobs,26 delayed results, the resignation of the chair of the Constitutional Council, and the assassina-

Electoral Violence in Sub-Saharan Africa

91

tion of the deputy chair of the Constitutional Council Babacar Séye.27 Nonetheless, the 1993 elections were considered by some to be more peaceful than the 1988 elections.28 Because of changes in the term of office for the president adopted as part of the previous reforms (tenure of office increased from 5 years to 7 years, limited to 2 terms), the 1998 elections were for the National Assembly only. They passed without incident. The 2000 elections, however, were a watershed moment in Senegal’s history as longtime opposition candidate Wade unseated the incumbent Diouf in the country’s most peaceful elections to date. Diouf won a plurality of the votes in the first round of elections held in February 2000, 41.3 percent to Wade’s 31 percent. But in the second round held in mid-March 2000, the opposition united to defeat Diouf and Wade won with 58 percent of the total vote. This election represented the first defeat for the Socialist Party/Senegalese Democratic Party since independence. Commentators hailed the 2000 election as a turning point for Senegalese democracy after Diouf accepted defeat and turned power over to Wade and the PDS. Once in power, however, Wade behaved much like his predecessors. He used political institutions as he saw fit. He introduced 25 constitutional amendments during his first 10 years in office, including a comprehensive overhaul as part of the 2001 constitutional referendum.29 He abolished and then reintroduced the Senate when he felt it was politically expedient to do so. He postponed elections in 2006, 2007, and 2009.30 During his first term in office, he appointed four different prime ministers including future opposition candidates Idrissa Seck and Macky Sall. Mbow (2008) characterized Wade’s presidency as “electoral authoritarianism” and decried Wade as a personalistic ruler with monarchical tendencies.31 She identified his frequent shuffling of ministerial portfolios—including the position of prime minister—his abuse of appointment powers over the Autonomous National Electoral Commission (CENA) and the Constitutional Court as well as the harassment of journalists and intellectuals as evidence of Wade’s attempt to consolidate his rule over Senegal.32 According to Kelly (2012), despite the hope that Wade’s presidency would usher in change, his leadership style was very similar to Senegal’s first president Senghor. Wade packed the Constitutional Court with allies, ignored the advice of CENA, delayed elections in order to ensure victory for his coalition, and censored journalists who wrote unfavorably on Wade and his son Karim.33 While there was a noticeable decrease in electoral violence between 1988 and 2009,34 there was also a contraction of opposition party operating space. The 2007 elections saw a major opposition boycott of legisla-

92

Senegal: Incidental Electoral Violence

tive elections. Galvan (2008) argued that, while the both presidential and legislative elections were “procedurally correct,” there was significant evidence of “heavy-handedness and a marked tendency toward personalism.”35 In addition, the opposition overestimated their chances of forcing a run-off in the first round of presidential elections and when Wade won a majority of votes (55.9 percent to second place candidate Idrissa Seck’s 14.9 percent), the opposition was sufficiently demoralized that they opted not to participate in the subsequent legislative elections. Their strategy was to regroup in order to effectively contest local elections in 2009. From 2008 to 2009, the opposition held a series of meetings around the country known as “Assises Nationales” in which they met with local leaders to discuss ways in which the opposition could unite and oppose Wade’s rule.36 In 2009 local elections the opposition coalesced under the banner “Benno Siggil Senegal” and was able to secure several key victories over Wade’s coalition, especially the indirect defeat of Karim Wade as mayor of Dakar.37 In 2011, the opposition was so confident of a 2012 victory that they “regularly suggest[ed] that any other outcome could only be produced by fraud.”38 Violence and the 2012 Elections

In hindsight, Senegal’s 2012 presidential election crisis was years in the making. Despite repeatedly assuring voters that he would only seek to serve two terms in office—and after introducing term limits in 2001 as part his comprehensive constitutional reform—Wade reversed course in 2009. Many commentators attributed Wade’s change in attitude to his failed attempts at installing son Karim as his successor. It was after the local and regional electoral losses in 2009 that Wade announced that he intended to run for a third term in office. He claimed that the two-term limit provision of the Constitution did not apply to him as it was enacted in 2001, a year after he began serving his first term in office. Term limits then did not include Wade’s 2000-2007 term in office and he should be cleared to run for another term. Wade was not the first African president to attempt to extend his rule beyond his original mandate, and he will likely not be the last. Executive term limits were widely embraced in the initial wave of democratic transitions in the 1990s and early 2000s. Many countries adopted such measures in an attempt to prevent the excesses of presidential power that seemed to frequently occur across the continent in the immediate post-colonial period. When ultimately faced with expiring terms in office, however, it has become all too common for Afri-

Electoral Violence in Sub-Saharan Africa

93

can presidents to attempt to remain in office. Presidents have challenged the applicability of term limits on technical grounds or sought to repeal them on ideological grounds, i.e., they limit vote choice. In most cases, civil society and opposition parties have protested such moves but their ability to prevent the consolidation of presidential power and the removal of term limits has been uneven. Presidents in Burkina Faso, Zambia, Nigeria, and Malawi have all tried—and failed—to amend their constitutions to remove executive term limits. Presidents in Uganda, Cameroon, Chad, and Niger, on the other hand, have been successful in repealing term limit provisions. The determinants of the success or failure of changes to executive term limits is currently the subject of ongoing research.39 Wade’s 2011 announcement of possible constitutional changes led to formation of several civil society oppositional groups including the M23 Movement and “Ne Touché Pas Ma Constitution.” “Y’en A Marre,” a collective of social activists and musicians that had formed in January 2011, allied with the other movements in order to protest Wade’s power grab. Indeed civil society was so organized that, according to one civil society leader, not a week went by without a major stakeholders meeting.40 In June 2011, Wade attempted to enact a flurry of reforms affecting political succession (creating the optional position of vice president) and changing the electoral threshold to win the presidency (reducing the requirement from 50 percent+1 to 25 percent). Protestors descended on the capital almost immediately. Because of the sustained protests of these groups—and after more than 100 protestors were injured—Wade retreated and rescinded the proposals, but he did not back down from his insistence that he was eligible to run for another term as president. Instead, he waited until January 2012 to officially announce his candidacy and relied on the Constitutional Court to favorably interpret the statute regarding presidential term limits. In the interim period between June 2011 and January 2012, the M23 Movement organized monthly protests and gatherings to maintain pressure on Wade.41 In December 2011, one of the main opposition candidates Idrissa Seck, who served as prime minister under Wade from 2002 to 2004, presaged the opposition’s response to the upcoming presidential election. He stated that if the Constitutional Court were to side in favor of Wade and his bid for a third term as president, that there would likely be violence; that he and his supporters would interpret such a ruling as a violation of the Constitution.42 He also said that, regardless of who were to come in second place in the presidential election, he would ally himself with the opposition in an eventual second round of voting.43 Other opposition leaders, such as Ousmane Dieng and Macky Sall, also reiterated Seck’s threat of violence if Wade were found eligible for an addi-

94

Senegal: Incidental Electoral Violence

tional term. Equally as vocal as the calls to violence and protest were the calls for peace. Religious leaders, trade unions, various representatives from civil society and international actors all issued statements leading up to the Constitutional Court’s decision appealing for peace ahead of the election. Wade formally submitted his candidacy for February’s presidential elections in January 2012 and awaited the Constitutional Court’s decision regarding term limits. The period between Wade’s announcement and the Court’s decision was tense. Thousands descended on Dakar as early as January 23rd to await the Court’s ruling. In anticipation of protests, the government moved to ban all demonstrations and gatherings in the capital city. After a sharp rebuke by the international community— and reportedly intervention by the European Union—the government rescinded its ban and instead dispatched riot police to Dakar. On January 27, 2012, the Constitutional Court upheld Wade’s interpretation of the statute and allowed his candidacy to proceed. Riots broke out the following day and continued up until the election on February 26th.44 There were reports of looting, arson, violent demonstrations, and multiple fatalities in the run up to the first round of the election.45 The opposition staged protests nearly every day, largely in Dakar but also in nearby suburbs and more remote towns such as Podor, Kaoulack, and Tivaouane. Protesters frequently clashed with security forces, resulting in several fatalities. Estimates place the number of deaths anywhere from 6 to 15.46 According to the Senegalese Red Cross, they treated 153 people for injuries related to pre-election violence during the month of February.47 The M23 alleges more than 500 were injured in pre-election violence.48 While not anywhere near the same magnitude of violence experienced in countries such as Kenya, Nigeria, or Zimbabwe, pre-election violence was significant and worrisome for Senegal. The first round of the presidential election saw Wade win a plurality of votes (35 percent) to Sall’s 27 percent but due to the 50 percent plus 1 majoritarian rules, a run-off was necessary. Wade’s failure to secure a first-round victory was a reflection of his growing unpopularity. He was even booed as he cast his own ballot on Election Day. It took him a few days during the vote tabulation process to concede outright victory, but eventually he announced that a run-off was indeed necessary. Interestingly, the day after the election, Wade promised that if he were to win the second round, he would form a government of national unity and include many members of the opposition. This type of political maneuvering is reminiscent of the politicking of his predecessor Diouf in the

Electoral Violence in Sub-Saharan Africa

95

1988 and 1993 elections, who offered to incorporate the opposition into government as a way of appeasing protestors. After the first round of elections, the violence stopped. There were no more protests or demonstrations. The opposition united behind candidate Macky Sall and delivered a second-round defeat to Wade. Sall received 66 percent of the popular vote to Wade’s 34 percent. The evening of the second round of voting, March 25th, Wade admitted defeat and conceded the election to Sall almost immediately.49 While seemingly precariously poised to descend into chaos, the country was able to resolve its electoral crisis through the ballot box and no further challenge was mounted. Given Wade’s attempt to groom his son as his successor, his insistence on running despite massive unrest and pressure to step down, and his available resources as incumbent president, it is surprising that he was willing to concede the election so easily. As such there are two related questions that need to be answered: what caused escalation of violence in a relatively peaceful country, and what defused the situation? In order to address these questions, it is important to examine the character of pre-election violence in Senegal in more depth. Unlike the electoral violence frequently found in countries such as Kenya, the violence in Senegal was not organized by political actors as part of larger campaign strategy but borne out of frustration and happenstance. In Ethiopia, Kenya, Nigeria, and Zimbabwe, for example, election violence is a strategy adopted by political actors to affect electoral outcomes. Politicians routinely hire gangs and militias to incite violence in order to intimidate and/or displace voters. The violence surrounding the 2012 election in Senegal can be characterized as “incidental,” meaning that the violence occurred as a result of cross-pressured actors who, because of exigent circumstances, were forced into different sides of the same political space. The vast majority of violence was a product of protesters engaging state security forces in disputes over the fairness of the electoral process. Protestors took to the streets to oppose what they perceived to be Wade’s attempt at a constitutional coup. Wade called up on state security forces to presumably maintain the peace but it is unclear what specific orders they were given. Some have accused Senegal’s security forces of using excessive force in their handling of protestors during this period.50 To reiterate, the violence that ensued was not meant to intimidate or displace voters ahead of the election but was a byproduct of two groups of agitated actors (opposition groups and state security forces) occupying a tense and fragile space.

96

Senegal: Incidental Electoral Violence

Because of the difference between incidental and strategic violence in terms of actors and intent, it is important to understand the factors that sufficiently incite and agitate actors to choose to place themselves in a vulnerable position where violence is a possible outcome. From the perspective of protesters, the government had not abided by the previously agreed upon rules of the game and thus had lost its credibility. The government could not be trusted to fairly and justly organize elections. Protesting before the election, regardless of the risk, was one of the only options left to prevent an outcome that many did not want and some genuinely feared: the re-election of Wade. The major triggers of the 2012 electoral violence in Senegal include: an attempted power grab by an unpopular incumbent politician; the belief that Wade would go to almost any end to secure entrenchment, as evidenced by his many attempts at manipulating institutions ahead of the elections; and pre-existing civic organizations prepared to mount a sustained challenge to such attempts. Senegal enjoys both a vibrant civil society and significant press freedoms, despite Wade’s attempts at censorship.51 Additionally, since at least the 1980s the opposition has steadily pressured successive governments to ensure free, fair, and transparent elections. As will be detailed below, there is little room now for Senegalese politicians to engage in widespread electoral fraud. Wade maneuvered (and some would argue manipulated) the institutions over which he had power. Wade had appointed all five members of the Supreme Court that ultimately ruled in his favor. But once Wade conceded to hold a second round of elections, as will be detailed below, he had no longer had any extra-judicial or extra-constitutional means to ensure victory. His fate was sealed and the opposition knew it. Election Management in Senegal

While there is a great degree of malleability in the political institutions in Senegal, there have been a series of reforms to ensure the autonomous and decentralized management of elections. From 1960 until 1998, election management was under the purview of the Ministry of the Interior. The opposition repeatedly called for autonomous and independent management of elections, alleging fraud and opacity of process; however, the ruling party did not seriously entertain this proposal until the late 1980s. Due to the electoral disputes of the 1980s, especially the violence surrounding the 1988 election, President Diouf established an electoral review commission in 1991.52 In 1992, the National Assembly adopted a new electoral code. Key provisions included: mandating the secret ballot, instituting two term limits for the presidency, guaranteeing polling

Electoral Violence in Sub-Saharan Africa

97

stations be monitored by all political parties, and a slight restructuring of the management of elections. 53 Despite these reforms, fraud occurred much as it had during the previous periods. After problems associated with local elections in 1996, a group of 19 opposition parties came together and called for the establishment of an independent electoral commission.54 After several rounds of discussions, Senegal’s first independent electoral oversight body was created: the National Elections Observatory (ONEL). ONEL oversaw the 1998 and 2000 elections, notable for the first defeat of a PS presidential candidate (Diouf) and the peaceful alternation of political power from the ruling party to the opposition. The ONEL, however, was only responsible for monitoring and supervising elections; the Minister of the Interior still retained responsibility for organizing and conducting elections. In the mid-2000s, there were again calls from opposition parties and civil society to enhance the independence of electoral management in Senegal in order to promote fair and transparent elections. The (new) ruling party preferred to strengthen the existing authority, the ONEL, while the opposition wanted to create a new independent permanent body. 55 After consultation with all political parties, in 2005 Wade and the National Assembly approved the establishment of CENA. CENA supervises elections and the Minister of the Interior conducts elections. The Directorate of General Elections, located within the Interior ministry, collects and maintains voter registration lists, produces the ballots, conducts civic education and awareness campaigns, and trains election officials, amongst other responsibilities. CENA is composed of 12 appointed members who serve staggered six-year renewable terms. They are appointed by the President with a limited amount of consultation with civil society, but there is no provision for removal of CENA members. Furthermore, CENA makes its decisions in secret. CENA is responsible for supervising all stages of the electoral process. It has the power to sanction government officials for electoral offenses and is a joint party to all legal proceedings related to electoral offenses.56 While CENA is still vulnerable to pressure from the executive, it exhibits a tremendous amount of autonomy compared to other countries in the region.57 It has full budgetary discretion and its budget is approved by the legislature, not the executive.58 Domestic groups view CENA as a partner to civil society and its independence is generally respected.59 In addition to the Ministry of the Interior and CENA, there are several other bodies that participate in Senegalese elections. These include: the Ministry of Foreign Affairs, the National Coalition for the Regulation of Broadcasting, and the courts.

98

Senegal: Incidental Electoral Violence

Domestic groups also play a large and important role in monitoring elections. In 2012, opposition parties dispatched monitors to all 11,000 polling stations.60 The domestic group, Réseau des Observateur Citoyens (Network of Citizen Observers, RESOCIT), a very broad network that comprised groups beyond traditional civil society, alone deployed 5,000 observers during the presidential and parliamentary elections.61 RESOCIT was an initiative of the Gorée Institute and the Collectif des Organisations de la Société Civile pour les Elections. It was established in March 2011. Additionally, the press provided live feeds of tabulation of results as they were reported.62 During the elections, RESOCIT established a “Situation Room” to create a technical platform to monitor elections, violence, and real-time results.63 Members of the political parties were invited to participate. An important task of the situation room was to develop scenarios and responses. For example, they had a response prepared if Wade would not concede. In short, there was little room in the electoral process to engage in manipulation or fraud. Because of this known reality, Wade attempted to exert his influence where he believed he could. He attempted to adjust electoral institutions well in advance of the elections to make victory easier for him to obtain. The 25 percent threshold for a first round victory could have been achieved with no fraud, as evidenced by his eventual showing in the first round. Lowering the threshold for a first-round victory was important because Wade knew as early as June 2011 that he would likely not obtain a majority of electoral support. There are reports that he was polling approximately 27 percent support at that point.64 Conclusions

The 2012 election campaign in Senegal was marred by protest, violence, and intimidation—all triggered by an attempted constitutional manipulation by incumbent president Abdoulaye Wade. His candidacy was announced despite legislation—which Wade himself supported in 2001— prohibiting an individual from serving more than two terms as president. Both the first and second rounds were tense but the month before the first round was particularly turbulent as protests, riots, and looting persisted for weeks on end. There were estimates of between 6 and 15 fatalities reported and hundreds injured as a direct result of the violence. Unlike much of the violence in Kenya, however, election violence in Senegal has rarely been strategic. The violence that manifested around the 2012 elections, in particular, was borne out of tense interactions between protesters and security forces. There is no evidence that the violence was pre-planned or meant to affect the outcome of a partic-

Electoral Violence in Sub-Saharan Africa

99

ular election. Instead, it was the product of frustrations that bubbled up over the electoral process and fears that elections would not be free and fair. Interestingly, the violence defused and de-escalated after the first round election results were respected by Wade, signaling to the opposition that further extra-judicial attempts to remain in power were unlikely. An examination of electoral management in Senegal and consequently the opportunities for fraud confirm that it would have been very difficult for Wade to have stolen the 2012 election. The types of “reforms” that Wade was forced to resort to, such as changing the electoral rules, and then relying on the Courts to rule in his favor, highlight the areas in which Wade felt he would be most influential and successful. They were also focused on events that would take place prior to Election Day. For the elections themselves, civil society was prepared. Groups were well mobilized and civil society was able to dispatch observers to every polling place in the country. The press was also able to provide real-time updates of election results. Taken together, these factors created an electoral environment in which the opposition was able to effectively vet the balloting process. Politicians and citizens alike believed they could trust the results because the process was inclusive and transparent. The Senegalese case is instructive as it underscores how transparent management and trusted observation can, by preventing electoral fraud, reduce tensions over the electoral process and promote peaceful transfers of power. However, the reform process in Senegal was long and drawn out, taking place over the course of decades. It took several iterations to identify the most appropriate reforms and to ensure the independence and autonomy of electoral management. Many reforms were followed by attempt to dilute them or undermine their efficacy. With such fluid institutions as is often found in newly democratizing countries, this is frequently the case. Thus, the Senegalese must remain vigilant in monitoring reforms and prepare for backsliding as politicians determined to access power adapt before the next elections.

Notes

                                                                                                            1

Nossiter, “An Atypical Unrest Troubles Senegal’s Election Season.” Alexander, “Youssou N’Dour Interview.” 3 Starr, “Senegal Gripped by Pre-Election Violence”; Labous, “Senegal Braces for Violence after Presidential Election.” 2

100

 

Senegal: Incidental Electoral Violence

4

“Macky Sall Senegal Election Win ‘Example for Africa.’” Galvan, “Political Turnover and Social Change in Senegal.” 6 Fall et al., “Senegal: Election Management Bodies in West Africa.” 7 Fall et al., “Senegal: Election Management Bodies in West Africa.” 8 Fall et al., “Senegal: Election Management Bodies in West Africa,” pp. 164-166. 9 O’Brien, “Political Opposition in Senegal: 1960-1967.” 10 Chafer, “France and Senegal” the End of the Affair?” 11 O’Brien, “Political Opposition in Senegal: 1960-1967.” 12 “African Elections Database.” 13 “African Elections Database.” 14 Stafford, “Senegal: May 1968, Africa’s Revolt.” 15 Fall et al., “Senegal: Election Management Bodies in West Africa,” pp. 164-166. 16 Creevey et al., “Party Politics and Different Paths to Democratic Transitions.” 17 Kante, “Senegal’s Empty Elections.” 18 Villalon, “Democratizing a (Quasi) Democracy.” 19 Creevey et al., “Party Politics and Different Paths to Democratic Transitions.” 20 Mbow, “Senegal: The Return of Personalism.” 21 Phillips, “Economic Austerity Remains a Divisive Issue in Senegal.” 22 Creevey et al., “Party Politics and Different Paths to Democratic Transitions.” 23 Kante, “Senegal’s Empty Elections”; Galvan, “Political Turnover and Social Change in Senegal.” 24 Johnson, “Will Senegal’s 2012 Presidential Election End in Violence?” 25 Kante, “Senegal’s Empty Elections.” 26 Galvan, “The Presidential and Parliamentary Elections in Senegal, February and June 2007.” 27 Kante, “Senegal’s Empty Elections.” 28 Kante, “Senegal’s Empty Elections.” 29 Fall et al., “Senegal: Election Management Bodies in West Africa.” 30 Kelly, “Senegal: What Will Turnover Bring?” 31 Mbow, “Senegal: The Return of Personalism.” 32 Mbow, “Senegal: The Return of Personalism.” 33 Kelly, “Senegal: What Will Turnover Bring?” 34 Fall et al., “Senegal: Election Management Bodies in West Africa.” 35 Galvan, “The Presidential and Parliamentary Elections in Senegal,” p.493. 36 Villalon, “Senegal: Assessing Risks to Stability.” 37 “Une Affair de Famille.” 38 Villalon, “Senegal: Assessing Risks to Stability,” p.12. 39 Posner and Young, “The Institutionalization of Political Power in Africa”; van Cranenburgh, “Democracy Promotion in Africa.” 40 Personal communication, March 2013. 41 “Senegal: Will the Momentum Last?” 42 Grange, “Senegal: Presidential Candidate Assesses Political Situation, Electoral Chances.” 5

Electoral Violence in Sub-Saharan Africa

 

101

43 Grange, “Senegal: Presidential Candidate Assesses Political Situation, Electoral Chances.” 44 Johnson, “Will Senegal’s 2012 Presidential Election End in Violence?” 45 Labous, “Senegalese Decry Pre-Election Violence”; “Senegal Violence Continues over President’s Third-Term Bid”; Djau, “3 Killed in Senegal Protests, Opposition Says.” 46 “Senegal Opposition Puts off Rally to Pay Tribute to Poll ‘Martyrs.’” 47 “Senegal Opposition Puts off Rally to Pay Tribute to Poll ‘Martyrs.’” 48 “Senegal Opposition Puts off Rally to Pay Tribute to Poll ‘Martyrs.’” 49 Nossiter,. “A Turbulence-Free Election in Senegal.” 50 “UN Concerned about ‘Excessive Force’ by Senegal Authorities.” 51 “Senegal: Freedom of the Press 2012.” 52 Fall et al., “Senegal: Election Management Bodies in West Africa.” 53 Fall et al., “Senegal: Election Management Bodies in West Africa.” 54 Fall et al., “Senegal: Election Management Bodies in West Africa.” 55 Fall et al., “Senegal: Election Management Bodies in West Africa.” 56 Fall et al., “Senegal: Election Management Bodies in West Africa.” 57 Fall et al., “Senegal: Election Management Bodies in West Africa.” 58 “Senegal: Independence Strengthened in a Mixed Model of Electoral Management.” 59 Personal communication, March 2013. 60 Kane, “Presidential Elections in Senegal: A Better Understanding of Recent Events.” 61 Personal communication, March 2013. 62 Kane, “Presidential Elections in Senegal: A Better Understanding of Recent Events.” 63 Personal communication, March 2013. 64 “Senegal’s Wade Drops Contested Electoral Reform.”

5 Liberia: Postconflict Electoral Violence

Liberia’s political history differs significantly from that of previous case studies Kenya and Senegal. It is one of the only countries in Africa to have never been formally colonized. Rather, Liberia began as a series of settlements founded by freed US slaves on the southern coast of West Africa. These settlements began to consolidate in the late 1830s due to antagonisms between the settlers and the indigenous groups that had already been living in the area. With mounting pressure from indigenous groups and the territorial expansion of the British colony in neighboring Sierra Leone, the settlers declared Liberia an independent state in 1847. Also unlike many other African countries, Liberia maintained regime stability throughout most of the 19th and 20th centuries until a military coup in 1980 threw the country into turmoil.. Prior to the coup, Liberia had held almost 50 presidential and legislative elections; however, competition was severely restricted and incumbents frequently ran unopposed. Furthermore, Liberia’s first 20 presidents were all from the Americo-Liberian political class and most were from one political party, the True Whig Party. Americo-Liberians, also referred to as Congos, have at most only ever comprised five percent of the total population (estimated at close to 4.4 million in 2014) but for approximately 133 years they held a tight grip on political and economic power in Liberia, effectively governing the country as an oligarchy and preventing the other sixteen or so ethnic groups from attaining power.1 This cleavage— the Americo-Liberians versus indigenous Liberians—was present at the founding of Liberia, has fueled two military coups and a brutish civil war, and remains politically salient in post-conflict Liberia. The 1980 military coup installed Master Sergeant Samuel K. Doe, an ethnic Krahn, as president. He was the first non-Congo to lead Liberia. During his rule, Doe allowed multiparty elections to take place in 103  

104

Liberia: Postconflict Electoral Violence

1985 but most Liberians believed these elections to have been rigged and the outcome completely fraudulent. Doe’s regime was considered corrupt and repressive by many. In 1989, Charles Taylor led a group of rebels into Liberia from neighboring Côte d’Ivoire in an attempt to depose Doe. This began the Liberian Civil War that lasted from 1989 to 2003, with a period of uneasy peace taking place for a few years in the late 1990s. In August 1996, a peace agreement was signed with all parties agreeing to hold elections in 1997. In a landslide, Taylor won the presidential election, receiving 75 percent of the vote. The elections themselves were peaceful but many argue there was an implicit threat of war looming over the country if Taylor were not the victor. During the war an estimated 250,000 persons were killed and upwards of 50 percent of the country’s population was displaced. The Liberian Civil War involved a multiplicity of actors including several insurgent groups from across Liberia; troop support from Sierra Leone, Guinea, Côte d’Ivoire Burkina Faso; and peacekeeping interventions from the Economic Community of West African States (ECOWAS) and the United Nations. In 2003, the warring parties finally agreed to a peace agreement and established a transitional government that would rule until general elections could be held no later than 2005. Many believe that Liberia is now on a path to recovery but how enduring this recovery remains to be seen. Since the end of the civil war, Liberia has held two successive elections, in October 2005 and October 2011. Election observers have deemed both elections to be free and fair but there is a palpable sentiment held by opposition supporters that both elections were somehow stolen from them. There have been significant allegations of fraud and an opposition boycott in 2011; however, electoral violence has been kept to a minimum and the immediate prospect of renewed conflict is low. Liberia’s lack of democratic institutionalization and the weakness of its political institutions, however, suggest that individual actors, if so inclined, could mobilize the country for widespread violence. The past two elections were organized and administered with significant assistance from the international community. Once the international community leaves and Liberia becomes wholly responsible for holding fair and peaceful elections on its own, the possibility of fraud and violence will likely return in full force. This chapter is an examination of elections in Liberia and how the quality of these elections speaks to the prospects for sustained peace and development in the postconflict context.

Electoral Violence in Sub-Saharan Africa

105

History of Modern Liberian Elections

After declaring independence in 1847, the political leadership of Liberia—comprised of former slaves from the United States—decided to model the new country after the United States, using the US Constitution as its template. The country’s first presidential election was held in September 1847. Joseph Jenkins Roberts of the Pro-Administration Party (PAP) defeated Samuel Benedict of the Anti-Administration Party (AAP). Every two years thereafter, elections were held until the president’s term was increased from two years to four years in 1907. Although Liberia began as a multiparty system and allowed for some party competition, by the late 1800s the True Whig Party had emerged as the dominant party. It won every subsequent election held in the 20th century until the coup in 1980. Elections were often rigged by the party and voting rights were limited to the elite political class until the universal franchise was introduced in the 1940s. After more than a century of repression by the Congo political elite, in1980, Samuel K. Doe and his followers staged a military takeover of the government and summarily executed then president William Tolbert Jr. and several members of his cabinet, all Americo-Liberian. After five years of autocratic rule, amid mounting domestic pressure for some level of liberalization, Doe allowed presidential and legislative elections to be held in October 1985. The official count declared Doe the winner with 51 percent of the vote but many believe that Jackson Doe (unrelated) actually won the election and that Samuel Doe’s National Democratic Party of Liberia (NDPL) engaged in large-scale fraud and vote rigging to “win” the election.2 In 1989 Charles G. Taylor led the National Patriotic Front of Liberia (NPFL) in rebellion against the government of Liberia in an attempt to remove Doe from power. The NPFL accused Doe of a variety of injustices including corruption and ethnic favoritism, claiming that he favored the Krahn over other indigenous groups such as the Gio and Mano. A breakaway faction of the NPFL, the Independent National Patriotic Front of Liberia (INPFL), succeeded in capturing Doe in September 1990. Infamous video footage of the execution shows warlord Prince Yormie Johnson, leader of the INPFL, participating in the brutal torture and execution of Samuel Doe.3 What began as an attempt by indigenous Liberians to wrest political control from Americo-Liberians in 1980 instead marked the beginning of a period of prolonged unrest, which eventually gave rise to the Liberian Civil War in 1989. The country descended into chaos and infighting with myriad attempts at resolution by the country’s neighbors, regional

106

Liberia: Postconflict Electoral Violence

bodies, and the international community. The Economic Community of West African States (ECOWAS) specifically, under the leadership of Nigeria, was heavily involved in the peace process. It established the Economic Community of West Africa Monitoring Group (ECOMOG) in 1990 in order to deploy peacekeepers to Liberia. After several failed attempts, the Abuja II peace accord was signed and implemented in 1996. As part of the agreement, elections were to be held within one year of the document’s signing. The 1997 elections were the keystone of the Abuja II peace accord, itself the 17th attempt at resolving the Liberian conflict.4 Given the situation on the ground after almost 8 years of intense fighting, the 1997 elections were problematic for many reasons. First, there were significant logistical issues that needed to be addressed such as integrating the sizable number of internally displaced persons (many Liberians refused to return until they were certain peace would hold) and overcoming the state of Liberia’s ruined infrastructure.5 At the time, there was no electricity, piped water, mail service, or reliable phone network in Monrovia and it was estimated that only 6 percent of Liberia’s roads were paved.6 The last census was undertaken in the 1980s and significantly out of date. Additionally, elections were scheduled during Liberia’s rainy season (May to October) in which many roads become impassable and peripheral communities become inaccessible. Nonetheless, ECOWAS and the international community pushed hard for Liberia to adhere to a quick timetable for the holding of elections. To address some of these logistical challenges, the electoral system was simplified. Instead of the plurality, single-member district system used for previous elections, a modified proportional representation system for legislative elections and simple majority rules for presidential office were adopted.7 Voters would cast one ballot for one party and the leader of the winning party would assume the presidency. Legislative seats would then be awarded on the basis of percentage of the vote the parties received. In theory, this would allow voters with little political information to participate in an important election; it would provide a clearly identifiable electoral mandate for the president and his or her party; and it would not require extensive voter education or political party strengthening programs. It was believed, or rather hoped, that the holding of elections for presidential and legislative offices would end the hostilities of the Liberian Civil War and return the country to peace. At the time, it was argued that elections would serve as a conflict resolution mechanism, providing a decisive “victory” to either Charles Taylor and the NPFL or one of the other main opposition groups. It was also believed that elec-

Electoral Violence in Sub-Saharan Africa

107

tions would provide legitimacy to the victor’s government, facilitating conciliation of the loser and cooperation between all participants. Despite protestations from the hastily assembled electoral commission, elections were held in July 1997. In hindsight, these glaring deficits and others led to the election being no more than a referendum on peace. Turnout of registered voters was high. Charles Taylor won handily; he and his newly formed political party the National Patriotic Party (NPP) garnered 75 percent of the vote to Ellen Johnson Sirleaf’s 9.5 percent.8 The rationale behind his support can be distilled down to its essence: a vote for Taylor was a vote for peace because of the tacit threat of the resumption of violence and civil war in the event that Charles Taylor did not win.9 Among Liberians, the saying was: “He killed my ma, he killed my pa, but I’ll vote for him anyway.”10 There was also a fear that Sirleaf was herself too politically and militarily weak to prevent renewed conflict with Taylor and the NPFL.11 The Carter Center and other observers deemed the vote to be free and fair. According to veteran reporter John O. Kollie, there was a definite implied threat of violence by the Taylor camp but even without the threat, Taylor was still a popular figure and would have likely polled around 60 percent of the vote in a truly free election.12 Unfortunately, peace was fragile and ultimately, short-lived. Amid allegations of severe mismanagement and repression by the Taylor regime, civil war resumed in 1999. Rebel groups, some supported by the regime in neighboring Guinea, resorted to large-scale violence in an attempt to remove Taylor from power. After mediation efforts by ECOWAS and international actors, another peace agreement was signed in June 2003. Shortly thereafter, Taylor resigned and fled to Nigeria. In August, ECOWAS sent a peacekeeping contingent to Liberia. The United Nations took over responsibility for internal security in October 2003. A transitional government was installed until elections could be held in October 2005. The Comprehensive Peace Agreement (CPA) of 2003 outlined the steps needed to take to de-arm and de-mobilize the main rebel groups and to prepare the country for multiparty elections. The 2005 Elections

As was established by the CPA, elections were held in October and November of 2005. Prior to elections, the transitional government adopted several amendments to the country’s previous electoral code and temporarily suspended other provisions, such as the requirement that a national census be conducted prior to the demarcation of constituencies. The National Elections Commission (NEC), an autonomous public commission,

108

Liberia: Postconflict Electoral Violence

was established to take over from the previously Electoral Commission (ECOM). Due to the complete lack of capacity of the Liberian state, still recovering from civil war, these elections were entirely managed by the international community. The United Nations and the African Union took up much of the responsibility for administering elections. The Carter Center and the National Democratic Institute (NDI), amongst others, engaged in civic education programs and both short-term and long-term electoral observation.13 The Carter Center and NDI, in particular, played a significant role in most every part of the electoral process.14 Twenty-two candidates competed for the presidency under majoritarian electoral rules in which the victor had to obtain at least 50 percent plus one of the total ballots cast. Seats in the House of Representatives were awarded based on simple plurality rules. Seats in the Senate were apportioned in a slightly more unorthodox fashion as voters were allowed to cast two ballots for the Senate and the top two vote getters in each constituency were awarded First Senate and Second Senate seats, respectively. The First Senator was to serve a nine-year term while the Second Senator was to serve a six-year term. This measure was undertaken to re-introduce staggered terms for the Senate The first round of elections gave George Weah, an ex-football star and founding member of the Congress for Democratic Change (CDC), a plurality of votes (28.3 percent). Ellen Johnson Sirleaf of the Unity Party, who had previously run in 1997 against Charles Taylor and had extensive experience working in international organizations, came in second with 19.8 percent of the vote. The major rebel groups from the second period of civil war (Liberians United for Reconciliation and Development, Movement for Democracy in Liberia) were largely absent from the election. The NPP fielded a candidate (Roland Massaquoi) but he received only 4.1 percent of the total vote. The legislative elections mirrored the fractured electoral landscape; the CDC received 16 percent of the vote for seats in the House of Representatives and the Senate while the Unity Party received a total of 12.5 percent of the vote for seats in the House and 13 percent of the vote for seats in the House. The remaining votes were distributed across some 20 parties. Independents won a total of 10 legislative seats out of a possible 94 seats in the House and Senate combined. Voter turnout was 74.9 percent. Because no candidate won a majority of the vote in the first round, a second round of voting was held on November 8, 2005. The secondround campaign was considered especially hard fought, due in no small part to the lack of an incumbent candidate, a rarity in the African setting.15 The lack of an incumbent candidate or political party is particu-

Electoral Violence in Sub-Saharan Africa

109

larly interesting because this also meant a relatively level playing field in terms of access to state resources and access to the media. Weah, despite or perhaps because of his popularity, was portrayed as uneducated and ill equipped to lead the country. He played football in the 1990s and 2000s for various European football clubs and was hugely popular in Liberia, especially amongst the urban youth population. Because of his football career, he was largely absent during the civil war. In addition, at the time of the 2005 election, he had not finished high school. His experience, or lack thereof, stood in stark contrast to Sirleaf, a former high-ranking official for the World Bank and the United Nations Development Program with a Master’s in Public Administration from Harvard University. Because of her international experience, she enjoyed broad international support but, domestically, she was less well regarded. Sirleaf was a part of the previous political elite having served in the cabinets of William Tolbert and Samuel K. Doe until Doe’s forces arrested her in 1985. It was alleged that she had ties to Charles Taylor and supported him in his early efforts to oust Doe.16 In fact, she has admitted to donating supplies and money to Taylor at the beginning of his military campaign but has since apologized for her “misjudgment.”17 In addition, although she is from Gola and Kru ancestry, some Liberians consider her to be from the Americo-Liberian former ruling class due to her early political career and her family’s close relations to several prominent Americo-Liberian families.18 Sirleaf won the second round of elections held on November 8, earning 59.4 percent of the vote to Weah’s 40.6 percent. Voter turnout decreased from the first round by about 14 percent, from 75 percent to 61 percent. While the elections were deemed free and fair by international observers including as the Carter Center, the European Union, and the African Union, the 2005 elections were not without problems. The main opposition parties levied allegations of fraud, mismanagement and vote buying.19 According to the CDC, in 2005 the number of voters was larger than the total population and a large number of ballots were secretly discarded in order to balance out the numbers.20 High-ranking members of the current Liberian government and opposition party supporters report that 2005 was not a fair election and that there were gross irregularities that the international community chose to ignore. 21 Even some staunch government supporters agree that “compromises were made” in 2005 in order to bring peace to the country.22 Despite all of these allegations, electoral violence was minimal. There were some outbursts in August during the general campaign and after the announcement of the second round of electoral results on November 23 but, overall, the elections were peaceful.23 Weah urged his

110

Liberia: Postconflict Electoral Violence

supporters to show restraint which some say helped to keep violence from taking hold of the country.24 Weah conceded in late December and Sirleaf was inaugurated in January 2006. She was Africa’s first elected woman president and has been the subject of much scholarly attention since.25 Sirleaf’s first term oversaw significant progress and reconstruction in Liberia. The country stabilized, the economy steadily grew and the poverty rate decreased from 64 percent to 56 percent during this period.26 As part of the long-term conflict resolution process, the CPA required that the Liberian government establish a Truth and Reconciliation Commission (TRC). The TRC was tasked with investigating war crimes committed throughout the civil war period. In February 2006, President Sirleaf swore the commission into office. In June 2009, the TRC issued its final report, which included the names of 52 people who should be barred from running for or holding public office for 30 years because of their associations with crimes and human rights violations that occurred during the civil war.27 President Sirleaf was on the list; however, in 2011, prior to the next elections, the Supreme Court invalidated the TRC recommendation barring Liberian politicians from seeking political office, claiming it violated due process.28 As of 2015, the many of the report’s recommendations had yet to be implemented. The Liberian Civil War is still a sensitive and difficult subject that Liberia is grappling with, but some resolution may have occurred as a result of the trial and conviction of former warlord Charles Taylor. Taylor was arrested by the International Criminal Court in 2006 and charged with war crimes in neighboring Sierra Leone. Taylor was found guilty in April 2012. The 2011 Elections

The country’s second postconflict general elections were held on October 11, 2011. The presidential election was a rematch between President Sirleaf of the UP and the CDC. George Weah, however, decided to run as vice-president and enlisted Winston Tubman, formerly of the National Democratic Party of Liberia (NDPL), to run as president on the CDC ticket. The CDC decided to combat what they perceived to be one of the major weaknesses of their unsuccessful 2005 presidential campaign— Weah’s lack of political experience—by assembling a ticket that included Weah and Tubman, a former ambassador to the United Nations and nephew of William Tubman, Liberia’s longest serving president. The logic behind this move was two-fold: to demonstrate that Weah was not power-hungry by his acceptance of the vice-presidential position, and to

Electoral Violence in Sub-Saharan Africa

111

show that the CDC ticket had international experience to rival that of Sirleaf.29 Additionally, in an attempt to counter Weah’s image as an uneducated footballer, he earned his bachelor’s degree from DeVry University in Miami, Florida in June 2011.30 With 71.6 percent of all registered voters participating in the first round, turnout was robust. Sirleaf received 43.9 percent of the first round vote to Tubman’s 32.7 percent. Prince Yormie Johnson, the senior senator from Nimba County and one of the videotaped torturers of Samuel K. Doe, came in third place with 11.6 percent of the vote. Legislative elections resulted in a very fragmented congress. Sirleaf’s Unity Party received 17.8 percent of the vote for the House of Representatives, winning 24 out of 73 available seats. The CDC received 12.8 percent of the vote resulting in 11 seats. The remaining seats—38 or 52 percent of the House of Representatives—were divided amongst 9 other parties. Nine independents won seats as well. Of the 15 Senate seats up for election, the Unity Party won 4, the National Patriotic Party won 4, and the CDC won 2. Four seats went to four additional parties and one independent was elected. Because no presidential candidate received more than 50 percent of the vote, a second round of elections was scheduled for November 8. Alleging massive voter fraud and an unfair playing field, on October 16 the CDC and several other opposition parties announced that they would be boycotting the second round of elections. Despite their boycott, Winston Tubman and George Weah were left on the ballot and the second round was held on November 8, 2011. Sirleaf was re-elected with 90.7 percent of the vote; Tubman received 9.3 percent. Turnout declined precipitously from the first round to 38.6 percent in the second round, a decrease of approximately 33 percent. Sirleaf was inaugurated for her second term in January 2012. Sirleaf campaigned on her accomplishments during her first term, including a massive debt forgiveness program in which Liberia was able to eliminate all of its external debt—estimated at $4.9 billion in 2007.31 During her 2005 campaign, Sirleaf vowed to be a single term president but in 2010 she changed her mind, stating that she was still had more work to do.32 Her campaign slogan in 2011 was “Monkey still working let baboon wait small,” a play on a common Liberian saying that inferred she, the monkey, was not yet done with Liberian politics.33 Again, President Sirleaf enjoyed considerable international support as evidenced by the Nobel Prize committee’s decision to award her the Nobel Peace Prize in early October 2011, days before the first round of the election.34 And again, Sirleaf was significantly less popular at home than abroad as demonstrated by the results of the 2011 election in which she

112

Liberia: Postconflict Electoral Violence

earned less than a majority of votes in the first round, forcing a second round. As in 2005, this election was anomalous in the African context. It was (and still is) highly unusual for an incumbent president in Africa to register such low levels of support. It is telling that Sirleaf was unable to manufacture a majority in the first round of elections; no incumbent wants to win office with an unpopular mandate. There are allegations that massive fraud took place before, during and after the campaign. Issues were raised regarding the composition of the NEC, the apportioning of electoral districts, the civic education campaign, the general campaign, the vote counting process, and the announcing of results. The following examines the allegations and opportunities the ruling party would have had to unduly influence the elections. Preparations began in January 2011 when the NEC started its voter registration drive. As per the Constitution of Liberia (Article 80 d), the NEC was required to redraw electoral constituencies after the 2008 census, the first conducted since 1984. The previous voter registration roll from the 2005 election was rendered obsolete by the updated census and, as such, the NEC spent January 10 to February 6, 2011 (extended once to February 12, 2011) registering the country’s estimated 2.1 million voters. The registration exercise was criticized by civil society as it was conducted after the delimitation of new constituencies and thus could lend itself to manipulation.35 In fact, this was a big problem according to several commentators because it allowed the government to effectively gerrymander districts.36 There were also significant challenges in accessing communities outside of the capital Monrovia for registration purposes. The exception to this could be found in some ruling party strongholds where, according to the Liberia Democratic Institute, the ruling party made sure these groups were accessed, in some cases using canoes.37 The International Foundation for Electoral Systems (IFES) worked closely with the NEC during the period before the election to provide civic education to the Liberian electorate. IFES partnered with several domestic civil society organizations to provide door-to-door civic education on the importance of registering and voting in the upcoming elections. They also made a Liberian-style short film that they showed to groups around the country.38 However, there were issues with the attempts made by the NEC at civic education, including lack of adequate funding and capacity and a lack of transparency, and perhaps impartiality, on the part of NEC when choosing which organizations to partner with as some groups were owned by NEC commissioners.39 It was also said that NEC provided only the most basic of information in its civic

Electoral Violence in Sub-Saharan Africa

113

education campaign (i.e., how to vote) and did little in the way of more complex or sophisticated education.40Some accused the Unity Party of “pre-campaigning” upwards of a year prior to the elections due to a series of billboards and radio and television spots touting the achievements of the government.41 There were also allegations that the political parties were attempting to move people out of their electoral districts prior to the vote to prevent them from voting. In February, the NEC announced that it was necessary to hold a national referendum prior to general elections to decide on four constitutional issues. The Liberian Congress approved them in 2010 but they required voter approval to enact. A national referendum was held on August 23, 2011. All four constitutional amendments (residency requirements for political candidates; retirement age for judges; timing of national elections; introduction of plurality rules for legislative and municipal elections) were defeated due to the constitutional stipulation that they each pass by a 2/3rds majority of voters; however, the Supreme Court ruled in September that amendment 4 (regarding the change from majoritarian rules to plurality rules for legislative and municipal elections) met that requirement if invalid votes were eliminated from the count.42 Both the holding of the referendum itself and the subsequent decision of the Supreme Court have been criticized. To some, the referendum was an unnecessary burden given the limited capacity of the NEC. Further, several stated that voters were confused as to the exact purpose of the referendum.43 The CDC told its supporters to abstain. Turnout for the referendum was around 34 percent. The 2011 elections were Liberia’s first self-managed elections since the 1980s.44 Unlike in 2005, Liberia and the newly constituted NEC were responsible for the conducting of the 2011 elections. The conduct and competency of the NEC became a flashpoint during the 2011 election after a controversy in announcing the first round results emerged. The Chairman of the NEC was forced to resign between the first and second rounds of the election and an interim Chair took his place. On October 25, 2011, Chairman of the NEC James M. Fromayan sent an official letter to the CDC that stated that the Unity Party received “394,370 or 32 percent of the valid votes cast” and “You [presumably the CDC] received the highest vote total among the 16 candidates thus qualifying you to participate in the run-off elections on November 8.”45 According to most reports, this was human error or a simple transposition of the results of the UP and CDC. The CDC, however, alleged that this is direct evidence of the manipulation of vote totals in the first round of the election.46 They claim it factored into their decision to boycott the second round of the elections. Shortly after the release of the letter, the

114

Liberia: Postconflict Electoral Violence

NEC Director of Public Information Bobby Livingstone was fired.47 In his own defense, Fromayan stated that he simply did not read carefully enough before signing the letter and that no fraud was perpetrated but, nonetheless, he resigned from office on October 30, 2011.48 The CoChair of the Commission, Elizabeth J. Nelson, took over his responsibilities and oversaw the second round of elections.49 By law, the board of commissioners of the NEC is to be appointed by the president of Liberia to a 6-year term, subject to Senate confirmation. President Sirleaf appointed the board of commissioners overseeing the 2011 election in June 2006. Some have argued that the president’s power to select and appoint the commission, with only a minor provision for congressional oversight, compromises the integrity and impartiality of the NEC. It has been alleged that Fromayan was a supporter of the Unity Party and a long-time friend of Ellen Johnson Sirleaf.50 Nelson’s husband, Blamoh Nelson of the Alliance for Peace and Democracy (APD), was the junior senator from Grand Kru County until he lost in the 2011 election to Peter Coleman of the CDC. In January 2012, President Sirleaf appointed Blamoh Nelson Internal Affairs Minister.51 In 2012 Chairman of the CDC George Solo described the 2011 elections as “one malfunction after another” and “an insulting and horrible process.” 52 The CDC alleged that bribery of political candidates and voter fraud, intimidation, and disenfranchisement were all rampant. Election workers intentionally marred votes by folding them prior to the ink drying, resulting in invalid votes. Ballot boxes were lost and/or unaccounted for by observers. CDC strongholds were told that they would not get electricity if they voted for the CDC. According to Solo, detailed election results were prematurely announced on privately-owned ratio stations with links to UP members, further evidence of the manipulation of the election by the UP with the collusion of the NEC. Finally, even though the CDC announced their boycott, the second round ballots were printed with Winston Tubman’s name on them. Solo argues this is because the NEC printed ballots far in advance and distributed them to polling centers upwards of a month in advance of the actual election.53 On 4 November 2011, the CDC officially announced that they would not participate in the second round. The belief that the 2nd round of elections would not be fair, says Solo, led directly to the decision to boycott the 2nd round.54 Additionally, many complained that the NEC did not adequately address opposition allegations of fraud and electoral mismanagement after the first round. Little was done in terms of dispute resolution. Jewel Howard-Taylor, former first lady of Liberia and NPP senior senator from Bong County in 2012, fully supported the boycott. According to her, the boycott was necessary because the opposition par-

Electoral Violence in Sub-Saharan Africa

115

ties felt collectively marginalized due to serious irregularities within the electoral process.55 Howard-Taylor, who chairs the Senate Committee on Public Autonomous Commissions and Agencies, was particularly critical of the dismissive attitude of the NEC who she claims failed to take opposition concerns and complaints seriously. They boycott, she claims, was intended to render the elections illegitimate. According to the opposition, the boycott was necessary to demonstrate to the voters and the international community that the elections were fundamentally unfair and to delegitimize the ruling party. The CDC insisted that in a fair election that it could have won the presidential election. Others, however, do not share the CDC’s assessment of their chances in the 2nd round. Some believe that the true motivation behind the CDC’s decision to boycott the second round of elections was, in fact, based on the party’s belief that it would lose in the second round and opted to not participate rather than accept electoral defeat. The party did not perform as well as it had expected to in the first round, its funding sources had dried up, and once Prince Johnson threw his support behind Sirleaf, the election was over.56 Thus while stated as an attempt to call into question the legitimacy of the electoral process, the boycott was, to some, the political machinations of a soon-to-be defeated party. The lack of resources and the low capacity of political parties and observers may have contributed to the belief that fraud occurred. According to Representative Gabriel B. Smith, chairman of the House Standing Committee on Elections and Inaugurations and an independent candidate in the 2011 elections, while political parties are constitutionally allowed to have representatives monitor elections at each polling site, most parties do not have the minimum requisite 4,457 observers to send, and parties assume there is fraud simply because they cannot ensure that there is not.57 Additionally, the chair of the Elections Coordinating Committee Oscar Bloh stated that while the committee observed no blatant instances of electoral fraud, there were technical errors that may have been misperceived by the opposition as fraud.58 There are many who have praised the management of the 2011 elections. According to the Carter Center Report, these were the best in Liberian history.59 ECOWAS, the EU, and the African Union, amongst others, certified these elections as free and fair. Richard S. Panton, former Secretary General of the Liberia People’s Party and current Deputy Director-General of the Liberia Institute of Public Administration, explicitly praised the transparency of the electoral process.60 Others do not share such a sanguine view of the electoral process. The Carter Center, in particular, was criticized by several actors for releasing its election assessment prior to the release of the official electoral results. Addition-

116

Liberia: Postconflict Electoral Violence

ally, international observers while present were few in number and, as such, logistically unable to observe activities at all polling stations. Search for Common Ground in Liberia chaired the Elections Coordinating Committee, a domestic electoral observation group fielded more than 2000 observers on the day of the first round of the election,61 but there were 4,457 polling places for each round of the election.62 Ushahidi Liberia, a non-profit tech company, collected crowdsourced information on the elections, including reports of fraud and violence.63 They reported relatively few incidents that would be considered electoral fraud or electoral violence. Of their total 5,676 incidence reports on the 2011 elections, 1,843 were categorized as “ECC Election Day Monitoring,” 1,780 as “Polling Stations,” 133 as “Campaigning,” and 93 as “Security Issues.”64 Ushahidi Liberia collected 68 reports of “Violence,” 68 reports of “Voter Issues,” and 12 reports of “Ballot Issues.” There are some limitations to the data collected by Ushahidi. Because it is based on voluntary citizen reports it may either underreport or overreport instance based on the ability of citizens to relay information with ease. The relatively low internet penetration in Liberia, which is concentrated in Monrovia, and the lack of cell phone infrastructure in the rest of Liberia suggest that fraud or violence that takes place outside of the capital city could go unreported. There are estimates that less than 0.5 percent of Liberians have access to the Internet.65 Cell phone usage is also quite low in Liberia, compared to the rest of the continent and world. This is expected to change in the near future but that does not help the reliability and validity of 2011 data. The pre-election violence that did take place included instances of vandalism, intimidation of Unity Party aspirants, and protests directed at CDC leadership.66 Despite the scant number of violent events that took place prior to and during the elections, there is a widely held belief in Liberia that there was the potential for significant electoral violence. The electoral environment leading up to the second round was very tense. There was one specific instance that many believe could have provided the spark needed for widespread violence to break out. On November 7, a riot broke out at an unsanctioned CDC protest when police and UN security forces blocked protesters from marching toward the capital, resulting in the death of at least one CDC supporter.67 The CDC refers to the incident as “Black Monday”. There was a sense that the CDC could have mobilized its supporters—mostly unemployed youths in the capital city—around this specific incident and yet the violence did not escalate. The second round of the election was held the next day on November 8. On November 15, it was announced that President Sirleaf had been re-

Electoral Violence in Sub-Saharan Africa

117

elected. While the opposition did complain and again alleged fraud, the outcome was ultimately accepted and the parties moved on. Although there were allegations of fraud and vote irregularity during the first round, a violent confrontation directly before the second round, and an opposition boycott of the second round, significant preelection or post-election violence did not take place. The incumbent had vowed previously not to run for a second term but reversed course as her first term was nearing its end. She was unpopular and seemed beatable. The election was projected to be close and many in the opposition were harboring grudges from the 2005 election, which they thought stolen from them. Both sides had sufficient motivation to use violence as part of an electoral strategy. Security forces and protesters came together on several occasions. Finally, political institutions were weak and malleable. The environment seemed ripe for serious election-related violence to take place but it did not manifest, begging the question: why? First, the country is not so far removed from the horrific violence of the civil war that wrecked the economy and caused mass suffering that the effects of violence have been forgotten. Many political leaders were directly impacted by the war. When there is a strong commitment by political leaders to avoid violence, political elites are unlikely to mobilize for violence before the election or to respond to the outcome with violence. CDC leadership argued that their supporters were angry and willing to fight but that they deliberately demobilized them in an effort to keep the peace. According to George Solo, there were numerous instances of irregularities on the first day of voting that would have incited violence were it not for the repeated calls by the CDC for their supporters and supporters of smaller parties to remain calm.68 President Sirleaf, recipient of the Nobel Peace Prize, was also unlikely to engage in violent tactics or reprisals. She repeatedly called for peace during the election season. The ruling Unity Party showed restraint in not responding to opposition with violence, despite the fact that they had the backing of the military and the police. In the absence of strong, autonomous institutions that are able to effectively manage and monitor elections, it then is left up to leadership and political elites to preserve the peace. If leaders have an aversion to using violence, then election violence will not take place but this commitment varies from individual to individual and context to context. And, unfortunately, it is unenforceable. Furthermore, this commitment is not always in place. In contrast to the post-election violence in Kenya in 2007 and the post-election violence in Nigeria in 2011, political elites were not urging their supporters to take to the streets to seek redress. The lack of post-

118

Liberia: Postconflict Electoral Violence

election violence in Liberia underscores the role that political elites, both from the ruling party and the opposition, can play in inflaming tensions or in allowing cooler heads to prevail. Second, there was a significant security presence in Liberia. According to the UN, on Election Day there were approximately 10,000 UNMIL total peacekeeping troops deployed around the country.69 To some, the international security presence was the key factor preventing the escalation of violence. Despite the opposition’s many grievances including the perception of fraud, violence simply could not have taken place with such a large presence of international peacekeepers watching. Incidental electoral violence, which often erupts as a result of interactions between partial or unprofessional state security forces and protesters, was unlikely because the peacekeeping forces were neither partial nor unprofessional. In addition, the international press devoted much attention to the country’s second postconflict election in which Africa’s first woman president (and Nobel peace prize recipient) was running for re-election. It was a unique and compelling story. Although the actual election observation and monitoring missions were lesser than in 2005, between the UN presence and the media attention, there was little opportunity to engage in strategic election violence without detection. In this case, the international community served to bolster and in some cases replace the roles played by government security forces, judicial actors, and domestic electoral observers. Conclusions

In 2005 after a prolonged civil war, Liberia held elections as a means of conflict resolution. Unlike in 1997 when elections were also held in a failed attempt to end civil war, the 2005 elections resulted in the brokering of a peace that has endured now 9 years and counting. In 2011, Liberians again went to the polls in the country’s second postconflict elections. Unfortunately, the past two Liberian elections have been beset with allegations fraud and vote rigging. Whether true or not, there is a palpable sentiment amongst the opposition that the 2005 and 2011 elections were given to Sirleaf. In isolated instances, violence has occurred but has not escalated due largely to an elite commitment to peace and a lack of opportunity to engage in violence without consequence. In both 2005 and 2011, there was a significant international presence that in effect took the place of the domestic institutions that are normally responsible for the quality and conduct of elections. After the international actors have departed, it will be up to the electoral commission, judiciary, civil society, and the press to monitor the behavior of electoral actors

Electoral Violence in Sub-Saharan Africa

119

before, during and after future elections. Any weaknesses or gaps in ability could provide actors not committed to peace the opportunity to engage in violence with impunity. Due to term limits, Sirleaf’s age, and the infighting of the Unity Party, some observers believe that it would be very difficult for George Weah and the CDC to lose the 2017 election in a fair fight. However, if they do lose, there are those that worry it could lead to significant violence. The sense on the ground is that the opposition will not be cheated a third time; in other words, a loss by the CDC in 2017 would amount to stealing the elections. The heightened significance of 2017 thus has the potential to destabilize the country significantly—especially if the CDC loses. It is clear that major reforms must be enacted prior to 2017. In the absence of strong institutions, which help to resolve inevitable domestic conflicts over the distribution of power and resources, peace relies on the commitment of individuals. It is impossible to predict how long this commitment to peaceful resolution of conflict will last and what will happen if effective monitoring and dispute resolution mechanisms are not in place prior to the next election. Coupled with the significant challenges still facing the country—unemployment is estimated at upwards of 70 percent of the population and economic growth has been recovering but still slow—the potential for violence in Liberia is ever present. Notes

                                                                                                            1

Harris, Civil War and Democracy in West Africa. Bøås, “Liberia—the Hellbound Heart?”; Lyons, Voting for Peace. 3 Howden, “Warlord to Kingmaker”. 4 Bekoe, “Toward a Theory of Peace Agreement Implementation.” 5 Lyons, Voting for Peace. 6 “World Development Indicators.” 7 Lyons, Voting for Peace. 8 “African Elections Database.” 9 Lyons, Voting for Peace. 10 Cooper, “On Day of Reckoning.” 11 Lyons, Voting for Peace. 12 Personal communication, July 2012. 13 “Observing Presidential and Legislative Elections in Liberia.” 14 “Observing Presidential and Legislative Elections in Liberia.” 15 For more on Africa’s strongman politics see Jackson, Robert H. and Carl G. Rosberg. “Personal Rule: Theory and Practice in Africa,” Comparative Politics 16, No. 4 (1984): 421-442 and Bienen, Henry. “Leaders, Violence, and 2

120

 

Liberia: Postconflict Electoral Violence

the Absence of Change in Africa,” Political Science Quarterly 108, No. 2 (1993): 271-282. 16 Harris, “Liberia 2005.” 17 Ford, “Sirleaf’s Nobel Award Fails to Improve Her Street Credibility.” 18 Bergner, “An Uncompromising Woman.” 19 Harris, “Liberia 2005.” 20 Personal communication, July 2012. 21 Personal communication, July 2012. 22 Personal communication, July 2012. 23 Sawyer, “Emerging Patterns in Liberia’s Post-Conflict Politics.” 24 Harris, “Liberia 2005.” 25 See Thomas, Gwynn and Melinda Adams. 2010. “Breaking the Final Glass Ceiling: The Influence of Gender in the Elections of Ellen Johnson-Sirleaf and Michelle Bachelet,” Journal of Women, Politics, and Policy 31, No 2 (2010): 105-131; Adams, Melinda. “Liberia’s Election of Ellen Johnson-Sirleaf and Women’s Executive Leadership in Africa,” Politics and Gender 4, No. 3 (2008): 475-484; and Bauer, Jacqui. “Women and the 2005 Election in Liberia,” Journal of Modern African Studies 47, No. 2 (2009): 193-211. 26 “Liberia Economic Outlook.” 27 “The Truth and Reconciliation Commission of Liberia: Final Report.” 28 Elmwood, “Liberia: No Justice, No Peace.” 29 Personal communication, July 2012. 30 “Liberia: George Weah Earns College Degree.” 31 McClanahan, “Africa’s ‘Iron Lady’.” 32 “Liberia President Ellen Johnson-Sirleaf to Stand Again.” 33 Keating, “Foreign Policy: Is Liberia’s Future in Her Hands?” 34 Ford, “Sirleaf’s Nobel Award Fails to Improve Her Street Credibility.” 35 Keating, “Foreign Policy: Is Liberia’s Future in Her Hands?” 36 Personal communication, July 2012. 37 Personal communication, July 2012. 38 Personal communication, July 2012. 39 Personal communication, July 2012. 40 Personal communication, July 2012. 41 Personal communication, July 2012. 42 “Liberia: Supreme Court Ratifies Constitutional Amendment.” 43 Personal communication, July 2012. 44 “Pre-Publication Final Report: National Elections in Liberia, Fall 2011.” 45 Personal communication, July 2012. 46 Personal Communication, July 2012. 47 “Former Liberian NEC Communications Chief Denies Being Opposition ‘Agent.’” 48 “President Sirleaf Accepts Resignation of NEC Chairman James Fromayan.” 49 “President Sirleaf Accepts Resignation of NEC Chairman James Fromayan.” 50 McLaughlin-Johnson, “Liberia ICAF Update Report.” 51 “‘I Need No Degree to Perform’: Blamo Nelson Hits Back at Critics.” 52 Personal communication, July 2012. 53 Personal communication, July 2012. 54 Personal communication, July 2012.

Electoral Violence in Sub-Saharan Africa

 

55

121

Personal communication, July 2012. Personal communication, July 2012. 57 Personal communication, July 2012. 58 Personal communication, July 2012. 59 “Pre-Publication Final Report: National Elections in Liberia, Fall 2011.” 60 Personal communication, July 2012. 61 Personal communication, July 2012. 62 Data from the National Elections Commission, Republic of Liberia. 63 Ushahidi Liberia used Internet and cell phone reporting to create crowdsourced maps of electoral related incidences which they then attempt to independently verify for accuracy. 64 “Liberia 2011: Monitoring the 2011 Elections in Liberia.” 65 “Internet World Stats.” 66 Genoway, “Liberia: Campaign Gets Deadly” and Daygbor and Genowaye, “Liberia: CDC Bloody Clash—Doe Others Flogged.” 67 “Liberia Election: CDC Monrovia Protest Turns Deadly.” 68 Personal communication, July 2012. 69 “UNMIL: United Nations Mission in Liberia Facts and Figures.” 56

6 Electoral Violence and Voter Turnout1

In theory, elections present a nonviolent alternative to conflict and instability, both of which have beleaguered much of sub-Saharan Africa in the post-colonial period. From 1960 to 1990, competition over state power in newly independent countries across Africa manifested in numerous coups and several brutal civil wars including those in Angola, Mozambique, Burundi, Côte d’Ivoire, and Uganda, to name a few. Across the continent, autocracy replaced autocracy many times over. Non-participatory regimes were the norm. Military dictatorships, personal autocracies and single party rule dominated the political landscape until the early 1990s when a combination of domestic and international pressures led many regimes to adopt multiparty elections. Today, only Eritrea has been seemingly immune to the global embrace of elections. The belief that popular participation and electoral accountability could address the conflict and instability plaguing the continent undergirded much of this pressure to change; however, in many cases, as has been detailed in the previous chapters, elections have brought with them conflict in the form of harassment, intimidation, and even death. Whereas previously much violence was a product of unconstitutional attempts at gaining power, now in many cases, violence seems to have become embedded in the electoral process. Part of the problem lies with the fact that the movement toward democracy has been slow and incomplete, often beginning and ending with the liberalization of electoral competition. In countries such as Gabon, Burundi, Democratic Republic of Congo, and Zimbabwe—all of which

                                                                                                           

This chapter is based on research and analysis conducted with the assistance of Dorina A. Bekoe. 123  

124

Electoral Violence and Voter Turnout

currently hold multiparty elections—significant repression and restrictions on freedom are still in place. Additionally, ruling parties are frequently able to manipulate the electoral landscape in their favor through the use of state resources including state-run media, creating an uneven playing field.1 Democracy does not exist without elections but elections can and often does take place without any assurance of democracy. Nonetheless, there is an enduring belief that elections will ultimately lead to a deepening of democracy. Electoral violence, however, seems to be an unabating trend. The belief in the “democratizing effect of elections” then must be reconciled with the persistence of electoral violence.2 It is unclear whether democracy can take hold and deepen while some voters are routinely exposed to violence and intimidation as a part of the electoral process. One way of approaching this seeming incompatibility is by examining the effects violent elections have on one of democracy’s main participants—voters. Although from a normative perspective we might believe that electoral violence is undesirable and harmful to democracy, it is still unknown exactly how electoral violence affects voter behaviors and attitudes. By examining the effects of electoral violence on individuals, upon whose consent democracy depends, we can gain a better understanding of the relationship between electoral violence and democratic development. How Electoral Violence Affects Voter Turnout

This chapter investigates the effects of electoral violence in Africa on voter turnout. The focus on voter turnout is important as among practitioners and scholars it is routinely seen as an indicator of political engagement, political enfranchisement, and democratic health.3 Indeed, low voter turnout has been cited as an indicator of low acceptance of democratic principles, of the failure of the elected officials to connect with constituents, and of the general disenfranchisement of the electorate. It is necessary to isolate this analysis to pre-election and day of election violence as post-election violence by definition takes place after an election and thus cannot affect turnout in a given electoral cycle. Violence that takes place before an election is meant to reduce competitiveness by intimidating or coercing behavior. Violence entrepreneurs may target political elites, electoral challengers, and /or the pool of potential or likely voters.

Electoral Violence in Sub-Saharan Africa

125

Because violence has the innate capacity to instill fear, it may seem intuitive that electoral violence would discourage political participation. When violence is directed specifically at voters, this could feasibly result in lower voter turnout. Indeed, there are certainly cases of countries where persistent electoral violence appears correlated with low voter turnout. In Nigeria’s 2011 parliamentary and presidential elections, voter turnout was 29 and 54 percent, respectively. In Zimbabwe, voter turnout in the 2008 parliamentary and presidential elections was 41 and 42 percent, respectively. In Comoros, the 2010 presidential election generated 52.8 percent turnout. Elections in all three countries are typically extremely violent, resulting in hundreds of fatalities. Conversely, there are countries where we should not be surprised to see high turnout, as elections are generally free, fair, and peaceful. These include Ghana, where the 2012 presidential and parliamentary election generated an 80 percent turnout; Botswana, where the 2009 parliamentary election resulted in a 74 percent turnout; and Mauritius, where 75 percent of the registered voters turned out for the 2010 parliamentary elections.4 But there are several surprising cases as well. Voter turnout in Mali, long lauded as a beacon of democracy before a March 2012 coup d’état, has not surpassed 38.2 percent since the start of its democratization process in 1996. Meanwhile, Ethiopia’s 2010 parliamentary elections saw a rise in voter turnout from 83 percent in 2005, its most violent election to date, to 93 percent in 2010.5 Understanding why notably peaceful elections can generate low voter turnout and notably violent elections can generate high voter turnout are among the puzzles this chapter seeks to address. But there are no studies—save two case studies on Nigeria and a dissertation written on Burundi discussed below—that systematically examine this question across Africa. Pre-election violence and day of violence are methods of directly affecting electoral outcomes by influencing vote choice and/or affecting the number of ballots cast. Post-election violence, on the other hand, occurs after the ballots have been counted. It may be a spontaneous reaction to the results or, as is often the case, it may be orchestrated to force some type of post-election bargaining in which electoral losers attempt to assert themselves into government through threats of more violence. These time periods are distinct in so far as they determine what outcome can be influenced by violence; however, violence may occur in one or more of these periods in any single election. Although the systematic study of the effects of electoral violence is a relatively recent phenomenon, there are two prominent case studies on Nigeria’s 2007 elections that explicitly examine the relationship between electoral violence and voter turnout. Michael Bratton measures

126

Electoral Violence and Voter Turnout

the impact of the threat of violence on the intention to vote in Nigeria’s 2007 election.6 He concludes these types of threats decreased the likelihood that individuals voted in the elections. Collier and Vicente (2012) investigate the effects of an international nongovernmental organization’s anti-violence campaign in Nigeria’s 2007 elections by comparing levels of voter turnout in the areas exposed to the campaign with those outside the program’s area7. Their study is particularly informative. Using a 2-week pre-election, anti-violence campaign by ActionAid International Nigeria (AAIN) in selected communities, they examine the effects of electoral violence on voter turnout in the communities inside and outside AAIN’s purview. Like Bratton, they find that the voters exposed to violence were less likely to vote. However, voters who were exposed to violence and were the targets of the AAIN anti-violence campaign were more likely to vote; voters who were affected by violence and were also exposed to AAIN’s campaign were 7 and 11 per cent more likely to turn out in the presidential and gubernatorial elections, respectively, than those only exposed to AAIN’s campaign.8 Violence, it seems, had a mobilizing effect on certain subset of voters. Similarly, Travaglianti (2014) examines the role of violence in the 2010 Burundian elections.9 According to her analysis, politicians in Burundi used violence explicitly to mobilize their own supporters. Still, these studies may reflect a particular environment and just one point in time. Beyond the research on Nigeria and Burundi, there are few studies that directly examine the impact of electoral violence on voter turnout cross-nationally; however, there are several researchers who make implicit assumptions about how electoral violence affects voter behavior in their broader analysis of electoral violence. Atubi (2008) and Meadow (2009), for example, both assert that violence decreases voter turnout, although neither provide any substantial evidence for this position.10 Hafner-Burton, Hyde, and Jablonski (2014) examine the timing of government-initiated electoral violence. In their analysis, they argue that violence may be used to prevent voters for the opposition from turning out, to “persuade” opposition voters for casting their ballots for the incumbent, or to “encourage” voters who would otherwise abstain to participate.11 Bhasin and Gandhi (2013) argue that governments are constrained in using violence against potential voters too close to an election as violence could adversely affect vote intention or participation.12 Although not directly measuring electoral violence, they find that governments are more likely to use repression against opposition elites directly before an election and civilian-oriented repression tactics in the periods after an election. This finding may indirectly support the

Electoral Violence in Sub-Saharan Africa

127

suppression thesis but their analysis is limited to presidential elections in autocratic regimes only so its generalizability is limited. Incidental violence could have a scare off effect where potential voters, especially if they are not a part of a protest movements or do not attend organized demonstrations, may prefer to keep themselves from perceived future harm by opting out of the process. A show of force by state security forces could have a deterrent effect on the turnout of potential voters. But it is also possible that incidental violence may reflect a particularly engaged opposition that is attempting to ensure fairness in the process. Incidental violence, although not directly related to the decision to vote, may instead be a reflection of mobilization efforts of an energized opposition. Strategic electoral violence is a means to an end: the winning of office. Violence as an electoral strategy is intended to ensure victory by reducing the competitiveness of an election. This can be achieved in multiple ways. High voter turnout of one’s own supporters and reduced turnout of opposition supporters both work to increase the likelihood of a candidate or party’s victory. The mobilizing aspect of electoral violence may at first seem surprising, but a review of how electoral violence has manifested in Kenya confirms that politicians have used violence to both discourage voters from participating and to intimidate voters into casting their ballot for a specific candidate or party. In the multiparty period beginning in 1992, violence has been employed in Kenyan elections for a variety of motivations, each based on a specific political and electoral context. In some cases, violence was a mechanism unleashed by political elites in order to suppress voters prior to election day; in others, it scared people into voting; and in still other cases, it punished vote choice after the fact. Moreover, violence has been used by different actors at different levels of government and at different times in the electoral process. In other words, the meaning of electoral violence depends on the perpetrators, the voters, the time in the electoral cycle, and domestic and international considerations. Additionally, Kenyan voters often seem to know what message is being sent and respond appropriately: they flee, ignore the message, or vote in large numbers. In sum, there is no consensus on the potential effects of electoral violence on voter behavior. Some argue that violence discourages political participation and some argue that electoral violence can either mobilize voters or coerce voters into casting their ballots for a certain candidate or party. Additionally, incidental and strategic violence might exert different effects on voter participation. And those who are organizing violence (i.e., the ruling party or the opposition) may have different and/or

128

Electoral Violence and Voter Turnout

conflicting methods and goals behind their use of electoral violence. Using the African Election Violence Database (AEVD) created by Straus and Taylor, it is possible to examine the relationship between electoral violence and voter turnout across sub-Saharan Africa.13 Because the Kenyan case has featured prominently in the analysis in this book, turnout figures from the 2002 and 2007 elections are also analyzed in order to further probe this relationship. Finally, survey data is utilized to examine individual-level self-reported patterns of behavior. Data Analysis

As was done in Chapter 2, Straus and Taylor’s initial four categories measuring electoral violence were collapsed into a three-part measure of electoral violence: no violence, incidental violence and strategic violence. From 1990 to 2014, the number of elections per country varies, depending on how many elections the country held during the time period. Some countries held only two elections during this time period (Democratic Republic of Congo) and others held as many as 10 elections (Comoros, São Tomé and Príncipe, and Togo). The data analyzed below includes a total of 289 observations (or elections) in 47 countries.14 Data on voter turnout and electoral institutions was collected from a variety of sources including the African Elections database, IFES, Lindberg’s (2006) Dataset on African Elections, the InterParliamentary Union database, and International IDEA.15 Legislative voter turnout (including both singular and concurrent elections for a total of 189 elections) and executive voter turnout (again, including both singular and concurrent elections for a total of 176 elections) are examined separately. Voter turnout in Africa for presidential and legislative elections averages 65.9 percent and 63.0 percent, respectively. Figure 6.1 reports average voter turnout by election type and incidence of electoral violence in sub-Saharan Africa. These data come from the pooled data set, which includes all elections in all countries with data available. The differences in average turnout between peaceful elections, elections that experience incidental violence, and elections that experience strategic violence are generally not statistically significant. These results also hold when comparing differences between nonviolent elections and both categories of election violence combined.

Electoral Violence in Sub-Saharan Africa

129

Percent

Figure 6.1: Voter Turnout and Pre-Electoral Violence in Africa, Registered Voters

100 90 80 70 60 50 40 30 20 10 0

No Violence Incidental Violence Strategic Violence

Executive Turnout

Legislative Turnout

The voter turnout figures reported above, however, are for the percentage of registered voters who cast their ballots in the election. It is possible that the decision to register to vote is also part of the broader electoral process that is affected by violence. Using turnout figures from the entire voting age population, and not just those who registered, a similar story emerges. There appears no significant difference in violence and voting age population turnout figures (See Table 6.1). Table 6.1: Voter Turnout and Pre-Electoral Violence in Africa, Voting Age Population Executive Turnout No Violence

Legislative Turnout

59.7%

57.7%

(n=64)

(n = 86)

Incidental Violence

56.4%

51.6%

(n = 69)

(n=63)

Strategic Violence

57.2%

50.2%

(n=29)

(n=28)

F(2,159) = 0.41

F(2, 174) = 2.00

(p=0.67)

(p=0.14)

F test=

130

Electoral Violence and Voter Turnout

Further complicating the data analysis is the fact that in this sample of cases some countries always have violent elections (Democratic Republic of Congo, Sudan) and some countries never have violent elections (Botswana, São Tomé and Príncipe). In these cases there is no variation in the key independent variable so absence/presence of violence cannot explain variation in voter turnout over time. One solution is to perform an isolated analysis of countries that do demonstrate variance in the absence or presence of electoral violence over time. This removes approximately 25 percent of the countries in the sample (10) and leaves data from 37 countries to examine. Table 6.3 reports voter turnout by type of election and occurrence of electoral violence for the isolated sample. While turnout is, on average, lower in legislative elections where violence occurs, the difference is not statistically significant. Based on descriptive analysis thus far, there appears to be no significant or consistent difference in voter turnout between violent elections and nonviolent elections. Table 6.2: Voter Turnout and Pre-Electoral Violence in Africa, Isolated Sample

No Violence Incidental Violence Strategic Violence F test=

Exec Turnout registered 61.6% (n=51) 66.4% (n=62) 65.5% (n=30) F(2, 140) = 1.43 (p=0.24)

Exec Turnout VAP 58.1% (n=44) 54.3% (n=66) 56.0% (n=30) F(2, 137) = 0.33 (p=0.72)

Leg Turnout registered 59.6% (n=61) 58.1% (n=63) 64.0% (n=32) F(2, 153) = 1.67 (p=0.19)

Leg Turnout VAP 52.8% (n=58) 48.8% (n=58) 50.2% (n=28) F(2,141) = 0.48 (p=0.62)

It may be the case that election violence has a lagged effect. Perhaps violence in a previous election deters undecided voters, non-partisans, and all but the most passionate voters from participating in subsequent elections. Unfortunately, because of the inconsistency and irregularity of elections in the larger dataset, it is difficult to test if violence from previous elections exhibits any discernable effect on turnout in future elections. Additionally, there are numerous cases of elections being interrupted by coups, conflict, and regime change which creates serious problems in utilizing the larger dataset to tease out longer-term effects. It is also difficult to use previous voter turnout as a consistent control across all cases. As such, a closer look at the Kenyan case over time al-

Electoral Violence in Sub-Saharan Africa

131

lows for more reliable longitudinal analysis. It also allows for the introduction of electoral competition into the analysis as Kenyan electoral constituencies remained the same for both the 2002 and 2007 elections. The Kenyan Case, 2007

While the 2007-2008 post-election violence has arguably garnered the most attention, pre-election violence has taken place more frequently than post-election violence in Kenya. Prior to the 2007 election 174 violent events occurred in the six months leading up to the election; 103 of them were related to election. Forty-three fatalities were reported. Using data from 2002 and 2007, it is possible to examine longer-term patterns in pre-election violence and turnout in the Kenyan case and also to explore the relationship between incidences of pre-election and postelection violence. For the purposes of the 2002 and 2007 elections, Kenya was administratively subdivided into 210 constituencies nested within 68 districts within 8 provinces. Data from the Armed Conflict Location Event Database (ACLED) reports, amongst other types of political violence, incidences of preelectoral violence that take place.16 This data is collected at the district level and the constituency level. Both constituency-level election data and district level data were analyzed in the hopes of capturing trends taking place at different administrative levels.17 Unfortunately, there is not sufficient information in the ACLED dataset to differentiate incidental electoral violence from strategic electoral violence; however, in the Kenyan case it is likely the case that much of the electoral violence was strategic based on anecdotal evidence in the form of investigative reports surrounding the elections.18 Additionally, and unlike in some other countries, electoral violence was perpetrated by both the ruling party and opposition parties. Figure 6.2 reports voter turnout in Kenya’s 2007 election based on the presence of pre-election violence. Of Kenya’s 210 constituencies, 64 (30.4 percent) had reports of violence before the 2007 election. Of Kenya’s 68 districts, 45.6 percent (31 districts) experienced some form of pre-election violence. In the 2007 election, average turnout for constituencies that experienced pre-election violence was 69.3 percent; in constituencies that did not experience pre-election violence, turnout was on average 70.6 percent. The difference is not statistically significant. At the district level, again, the difference is not statistically significant.

132

Electoral Violence and Voter Turnout

Figure 6.2: Voter Turnout and Pre-Electoral Violence in Kenya, 2007

100 90 80 Percent

70 60

Violence before the election

50 40

No violence before the election

30 20 10 0 Constituency Level Turnout

District Level Turnout

Kenyan elections between 1997 and 2007, however, were conducted using the same constituencies and without major reform thus it is possible to examine if historical patterns of electoral violence are related to future voting patterns. Using ACLED data from the 6-month period before the 1997 and 2002 elections (corresponding to the Straus and Taylor time line19), it is possible to examine turnout patterns in districts that had a history of election violence before either of the previous elections.20 Of Kenya’s 68 districts, 35.3 percent (24) had no previous experience with election related violence before the 2007 polls. Both the 1997 and 2002 elections took place within these districts—also without violence. For districts that had a history of election violence before the 2007 election, voter turnout in 2007 was 68.9 percent; the 24 districts that did not have a history of pre-election violence averaged 72.4 percent voter turnout. This difference is also not statistically significant. In the aggregate, looking specifically at country-level and subnational-level data, it is difficult to detect a significant and consistent relationship between electoral violence and voter turnout. It might be that because of the variety of types and purposes of electoral violence, the net effect of electoral violence is difficult to detect. The relationship between fraud and violence may also be obscuring the effects on voter turnout. As electoral fraud is often times used as a supplementary strategy or violence is resorted to when attempts at fraud are unsuccessful, it

Electoral Violence in Sub-Saharan Africa

133

could be the case that voter turnout is an inherently flawed metric with which to attempt to capture the effects of electoral violence. Willingness to Vote in the Future

Utilizing data from individual-level surveys may be a more appropriate method of examining the effects of electoral violence on citizens. Survey data allows individuals to self-report relationships and takes into account individual-level perceptions. Analyzing survey data also mitigates some of the potential influences of fraud, electoral mismanagement, and the willful misreporting of election data. Official turnout figures may reflect some or all of these factors. Finally, the use of survey data allows us to include, albeit implicitly, the decision to register to vote. A drawback, however, is that survey data does not allow for a more nuanced look at the effects of different types of electoral institutions and different type/timings of electoral violence on individuals. Round 4 of the Afrobarometer survey was conducted in 20 African countries between 2008 and 2009.21 While covering almost half of the countries included in the AEVD, there is a substantial difference between countries surveyed and countries not surveyed. In terms of openness and democratic institutions, the 20 countries included in the survey perform much better than the 28 sub-Saharan African countries in which surveys were not conducted. For example, the average Polity score at the time of the survey for countries included in Afrobarometer Round 4 is 5.15 (moderately democratic). For the remaining 28 countries in subSaharan Africa not included in the survey, the average Polity score was 0.7 (marginally autocratic). Because of this substantial difference, the generalizability of the following data analysis must be limited to the countries in the sample and not the whole of sub-Saharan Africa. Based on the previous analysis in this book, it is likely that the relationship between election violence and voter behavior is mediated by the openness of the elections and the ability of voters and aspirants to participate. Nonetheless, some striking patterns emerge when analyzing how electoral violence affects voter turnout in more the democratic and open countries of sub-Saharan Africa.

134

Electoral Violence and Voter Turnout

Table 6.3: Fear of Electoral Violence in Africa

Do not fear EV A little bit fear of EV Somewhat fear of EV A lot fear of EV

Afrobarometer countries (20) 44.4%

Kenya (2008)

Senegal (2008)

Liberia (2008)

17.5%

58.3%

42.4%

21.5%

20.6%

14.2%

31.7%

15.1%

24.2%

13.8%

14.5%

19.0%

37.8%

13.7%

11.3%

Afrobarometer Round 4, 2008 – 2009

Round 4 of the Afrobarometer survey specifically asks individuals “[d]uring election campaigns in this country, how much do you personally fear becoming a victim of political intimidation or violence?”22 The coded options allow the respondent to choose: a lot, somewhat, a little bit, or not at all. Table 6.3 summarizes the self-reported distribution of fear of electoral violence in all 20 countries included in the Afrobarometer Round 4 survey alongside the book’s three case studies, Senegal, Kenya, and Liberia. A slight majority of total respondents in the Afrobarometer sample stated that they feared electoral violence to some degree (55.6 percent) whereas 44.4 percent stated that they did not fear electoral violence at all. Of the full sample, nearly one-fifth reported fearing electoral violence “a lot.” Across the three cases, there is significant variation in fear of election violence. In 2008, a majority of Senegalese (58 percent) stated that they did not fear electoral violence at all. In Liberia in 2008, three years after the country’s first post-war election, the level is slightly lower (42.4 percent) but a plurality of Liberians still reported not fearing electoral violence. Kenyans, on the other hand, reported significant levels of fear when the survey was conducted—less than 1 year after the postelection violence. Only 17.5 percent of Kenyans claimed they did not fear electoral violence; more than 80 percent reported some level of fear of election-related violence. Round 4 of the Afrobarometer survey also asks individuals which party, if any, they planned to vote for in their country’s next elections. Of respondents in the full sample, 69.2 percent responded in the affirmative that they would vote in the next election and provided the specific party or candidate that they intended to vote for. The remaining 30.8

Electoral Violence in Sub-Saharan Africa

135

percent stated that they would not vote, that did not know if they would vote or not, or they refused to answer the question. Roughly eight percent of those surveyed stated in absolute terms that they would not vote in the next election. Ten percent refused to answer the question and 14 percent stated that they did not know. It is impossible to know with any certainty the true vote intentions of respondents who either refused to answer the question or stated that they did not know who they would vote for. There are multiple ways of treating this type of question and response. Those who don’t know or refuse to answer the question could be considered uncommitted voters. Their lack of commitment to the process could allow for these groups of respondents to be incorporated into a “do not plan on voting” category. Some opt to treat don’t knows and refusals in surveys as missing data and remove these individuals from analysis all together. However, removing these individuals from the survey may systematically bias the results. Those who refuse to answer the question or genuinely do not know how they intend to act in the future might also be affected by election violence. In fact, violence may be influencing their agnosticism or their decision to not disclose this information. Electoral violence, or the belief that it might take place in future elections, could introduce uncertainty into the decision-making process. I have chosen retain all four responses and treat them as distinct categories because I believe this information to be valuable, albeit inconclusive. Table 6.4: Willingness to Vote in the Future

Will Not Vote Will Vote Refused to Answer Don’t Know

Afrobarometer Countries (20) 8.0%

Kenya (2008)

Senegal (2008)

Liberia (2008)

10.1%

5.2%

15.2%

67.6%

74.2%

55.0%

55.2%

10.1%

4.5%

13.5%

6.0%

14.2%

11.1%

26.2%

23.5%

Afrobarometer Round 4, 2008 – 2009

The distribution of my key dependent variable is found in Table 6.4. Even after the 2008 post-election violence, Kenyans were above the

136

Electoral Violence and Voter Turnout

sample average in terms of reporting that they planned to vote in the future: 74.2 percent of those surveyed stated that they would vote in the next election. If this number seems high, in comparison, voter turnout for the 2013 elections was reported to be 85.9 percent. However, when adjusted for the size of the population eligible to vote versus those who actually registered, voter turnout is estimated to be closer to 55.6 percent.23 In Senegal and Liberia, the percent surveyed who stated that they absolutely would vote in the future was 55 percent and 55.2 percent, respectively. Almost one-quarter of those surveyed in both countries stated that they did not know if they would vote in the next election. Both countries generally report lower levels of voter turnout than Kenya. In the 2007 Senegalese presidential election, turnout of registered voters was 70.6 percent, but that is the highest turnout has been since prior to 1983. Most presidential elections in Senegal generate turnout levels between 50 and 60 percent. Parliamentary elections typically report lower overall levels of voter turnout in Senegal. For the 2007 parliamentary election, turnout was 34.7 percent. Turnout in Liberia’s 2005 elections, the first held since the end of the Liberian Civil War, was 61 and 77 percent for presidential and parliamentary elections, respectively. Table 6.5 reports the bivariate statistics for the relationship between willingness to vote in the future and fear of violence for the full Afrobarometer sample, and compares this to levels found in Kenya, Senegal, and Liberia. The mean values for will vote in the future are compared for those who fear electoral violence and those who do not. In the Afrobarometer sample, the relationship between fear of election violence and willingness to vote in future elections is small but significant. Those who fear electoral violence were approximately 3 percent less likely to state that they would vote in the future election than those who reported no such fear. In Kenya, there seems to be no significant relationship between fear of election violence and self-reported intention to participate in future elections. This may be a function of the number of different groups engaging in electoral violence prior to the 2007 election. Both the ruling party and the main opposition parties were responsible for organizing violence. The Kenya National Commission on Human Rights affirmed that a number of different ethnically-based illegal gangs (Mungiki, Luo Taliban, Siafu, and Bukhungu) had been mobilized before the election to act at the behest of the parties.24 Many of these same groups had been used prior to the 1992 and 1997 elections as well.25

Electoral Violence in Sub-Saharan Africa

137

Table 6.5: Fear of Electoral Violence and Willingness to Vote

Fear No Fear Chisquare

Afrobarometer (2008 – 2009) 66.4% 69.3% Chi-square = 25.5 (p= 0.00)

Kenya (2008)

Senegal (2008)

Liberia (2008)

74.8% 71.6%

58.0% 52.8%

49.2% 63.7%

Chi-square = 0.9 (p=0.35)

Chi-square = 25.0 (p=0.00)

Chi-square = 3.3 (p=0.07)

Afrobarometer Round 4, 2008 – 2009

In Liberia and Senegal, there appears to be a strong relationship between fear of election violence and willingness to vote in future elections, but in opposite directions. Of Liberians who stated that they feared electoral violence, less than half (49.2 percent) responded that they would definitely vote in the future. Those who stated that they did not fear election violence were much more likely to state that they would vote in future elections—63.7 percent. In Senegal, those who feared electoral violence were more likely to state that they would vote in the future than those who reported having no such fear (58 percent versus 52.8 percent). This may be reflective of the nature and primary agents of election violence in Senegal. The violence that had broken out prior to the 2007 elections was incidental in nature and sparked by opposition protests against President Wade’s attempts at unduly influencing the electoral process through strategic delays of the parliamentary election. In order to take this analysis further, I ran a series of logistic regressions to estimate the impact of fear of electoral violence on an indivdual’s stated likelihood of participating in future elections for the full sample of countries. I retained the nominal treatment of intention to vote because there is valuable information to be gleaned from all four categories of respondents. I ran four separate logistic regressions estimating each outcome (will not vote, will vote, refused to answer, don’t know). Fear of electoral violence is my key independent variable. It is measured as an ordinal variable where 1 corresponds to no fear of violence and 4 corresponds to a lot of fear of election violence. Partisanship and past vote behavior are strong predictors of an individual’s likelihood of voting in the future so I include control variables

138

Electoral Violence and Voter Turnout

for each.26 Partisanship is measured with two dummy variables, one for ruling party supporters and one for opposition party supporter. The third category, non-partisans, is the remaining respondents not represented by either dummy variable. Based on my understanding of electoral violence, I believe fear of violence could have different effects on ruling party supporters, opposition party supporters, and non-partisans. Violence may spur ruling party and opposition supports to action and, at the same time, alienate non-partisans. Partisans in general might be more exposed to violence due to their level of political engagement. Those who identify as non-partisans might be less exposed to or aware of violence accompanying political campaigns. Thus, I include variables to capture possible interaction effects between self-reported partisanship and electoral violence. Past vote is a binary variable where 1 corresponds to individuals who state they voted in their country’s most recent election and 0 corresponds to those who say, for whatever reason, they did not vote. Turning to the literature on voter turnout in Africa specifically, I relied on the foundational work of Michelle Kuenzi and Gina Lambright (2007). They find that in sub-Saharan Africa, voters in rural areas tend to turn out in higher numbers than their urban counterparts, contrary to the findings of many studies conducted in Western countries. Thus, I include controls for urban/rural location in addition to the other standard socioeconomic variables that are generally believed to affect voter turnout (age, gender, and educational attainment).27 Age is a ratio level variable, gender is binary measure, and educational attainment is also measured as an ordinal variable ranging from 0 for no formal education to 10 for post-graduate education. The dataset includes responses from approximately 1,200 individuals from each of 20 African countries for a total of more than 26,000 respondents. I include random effects, most appropriate for the hierarchical nature of the dataset which combines individual responses from multiple different countries. Random effects also allow us to implicitly include country-specific information so as to more precisely estimate the relationships between variables.28 Table 6.6 reports the coefficients for the four logistic regressions estimating the effect of fear of electoral violence on individual-level vote intention. Fear of electoral violence is significantly and consistently related to future vote intention across the four models but the direction of the effect differs based on individual-level partisanship. Because willingness to vote in the future is most directly related to measures of voter turnout, the remaining section will look at this relationship more closely.

Electoral Violence in Sub-Saharan Africa

139

Table 6.6: Fear of Electoral Violence and Willingness to Vote in Africa, Random Effects Logit Regressions Will Not Vote

Will Vote

Refused to

Don’t Know

Answer Fear of EV

.05**

-.02

.03

-.03

(.02)

(.02)

(.03)

(.02)

Ruling Party

-2.9***

2.91***

-2.08***

-2.17***

Supporter

(.21)

(.09)

(.16)

(.12)

Fear X Ruling

.006

-.004

-.04

.05

Party Supporter

(.09)

(.04)

(.07)

(.05)

Opposition Party

-1.99***

1.07***

.23**

-.84***

Supporter

(.15)

(.07)

(.09)

(.10)

Fear X

.04

.1***

-.12**

-.12**

Opposition Party

(.06)

(.03)

(.04)

(.04)

Voted

-.65***

-21***

.10**

.02

Previously

(.05)

(.03)

(.05)

(.04)

Supporter

Urban

-.11*

-.06*

-.03

.17***

Residence

(.05)

(.03)

(.05)

(.04)

Age

.008***

-.01***

.01***

.006***

(.002)

(.001)

(.002)

(.001)

.08

-.16***

-.02

.22***

(.05)

(.03)

(.04)

(.04)

.08***

.004

.02*

-.07***

(.01)

(.01)

(.01)

(.01)

-2.05

.37

-2.51

-1.84

Gender

Education

Constant

140

Electoral Violence and Voter Turnout

Log Likelihood

-5718.89

-12948.97

-7556.34

-8615.01

sigma

.68

.59

.65

.92

rho

.12

.09

.12

.20

n

26,638

26,638

26,638

26,638

Afrobarometer Round 4, 2008 – 2009 Standard Errors in parentheses. *p=.05 **p=.01 ***p=.001

In estimating the likelihood that an individual unequivocally states that they will vote in the next election, fear has no effect on ruling party supporters, but a positive effect on opposition party supporters and nonpartisans. For the full sample of cases, fear appears to have a mobilizing effect on opposition members and non-partisans; however, it is unclear who these voters intend to cast their ballots for in the future. It seems plausible that those who identify as opposition members and/or nonpartisans in the survey and fear electoral violence might chose to vote for an opposition or independent candidate in the future as a result. I ran the same logit regressions for Kenya, Senegal and Liberia that were used for the full Afrobarometer sample.29 Table 6.7 compares the directionality of the marginal effects of electoral violence on willingness to vote in the future by partisanship for all the Afrobarometer countries with Kenya, Senegal, and Liberia broken out. Marginal effects are calculated as the difference in predicted probabilities of willingness to vote for respondents if they state they fear electoral violence “a lot” and if they state that they do not fear it at all. Predicted probabilities were estimated holding all control variables at their median or mean values. Estimates are reported in parentheses but are more useful as a reference and not a precise figure.

Electoral Violence in Sub-Saharan Africa

141

Table 6.7: Impact of Fear of Electoral Violence on Willingness to Vote in the Future in Kenya, Senegal, and Liberia Afrobarometer

Kenya

Senegal

Liberia

Ruling Party

None

Opposition

Increase (6%) Increase (8%)

Increase (16%) Increase (6%) Decrease (13%)

Decrease (6%) Increase (15%) Increase (4%)*

Decrease (9%) Decrease (26%) Decrease (20%)

Non Partisans

*Marginally significant at p=.01 level

In Kenya, fear of electoral violence in 2008 was related to an increased willingness to vote in the future for ruling party and opposition supporters. Non-partisans were less likely to state that they would vote in the future if they feared electoral violence. In the context of the 2007 election, violence might have simultaneously mobilized partisans of all stripes while alienating those who did not identify with any particular party. This finding could be a product of the identities of the main agitators of the election violence that occurred in 2007 and the messages that they were spreading. Fear of an electoral loss featured prominently in the campaign discourse and there is ample evidence that both the ruling party and main opposition party engaged in electoral violence. The ruling party and main opposition party both stoked fears of ethnically-based violence amongst their own supporters if they were to lose. This finding might also reflect how directly respondents had been affected by electoral violence given the recentness of the PEV in Kenya to the survey and how intense and pervasive it was. In Senegal, fear of electoral violence was related to a lower probability of ruling party supporters voting, but much higher probabilities of voting for opposition party supporters and marginally higher probabilities for non-partisans. Incidental violence could be mobilizing the opposition (and non-partisans to some extent) and slightly deterring ruling party supporters from future participation. In Liberia, fear of electoral violence is associated with significantly lower levels of willingness to vote for all those surveyed, but the effects are more pronounced for opposition supporters and nonpartisans. This could be a function of the newness of the political parties and the shallowness of partisan attachments. Violence, even the threat of it, might discourage individuals from participating in

142

Electoral Violence and Voter Turnout

elections in postconflict environments. In sum, there is evidence that fear of electoral violence does affect voter participation but the effects are context-specific and vary based on partisan attachment. Generally speaking, non-partisans are less likely to vote if they fear electoral violence whereas partisans are more likely to vote if they fear electoral violence. The causal mechanism at this point, however, has not been adequately identified. As a robustness check, I ran separate logit models for each subset of countries based on the type of election violence that occurred in the most recent elections prior to the Round 4 survey (none, incidental, and strategic). For the countries in the sample that did not experience violence during their recent elections (Benin, Botswana, Burkina Faso, Cape Verde, Lesotho, Madagascar, Mali, Namibia, and Zambia), fear of electoral violence was not significantly related to willingness to vote in the future. For the eight countries that experienced incidental strategic violence, fear of election violence was significantly related to increased willingness to vote in the future on the part of the opposition and marginally so for non-partisans. For the three countries that experienced strategic violence, effects on partisans were more complicated (opposition supporters in Nigeria were less likely to state that they would vote in the future if they feared election violence whereas opposition supporters in Zimbabwe were more likely to state they would vote in the future if they feared violence). Non-partisans in Nigeria and Zimbabwe were both more likely to state that they would vote in the future if they were afraid of election violence. There is evidence to suggest that violence affects voter participation. Violence can serve to both mobilize and suppress voter participation, even in the same election. Individual-level partisan attachments are the pivotal factor which determines if violence increases or decreases future voter participation. Additionally, the specific type of election violence that occurs and the message that it is meant to convey is extremely important in determining how violence affects voter behavior. Based on this survey data alone, it is not possible to determine if electoral violence affects or somehow changes partisan attachments or candidate evaluations, something that future research should examine. It must be noted that this finding applies only to the self-reported intention to act in the future and might not extend to the behavior itself. There is no guarantee that those who state they will vote in the future will actually follow through on this intention, or that fraud will not nullify the effects of this relationship when examining electoral outcomes. Nonetheless, electoral violence

Electoral Violence in Sub-Saharan Africa

143

does appear to have an effect on voters, and potentially, their behavior. Before these findings can be generalized any further, it is important to remember that elections have multiple purposes and meanings depending on the political environment in which they are held. Democratic and autocratic elections differ substantially in terms of how free individuals are to participate in voting and whether voters believe that their vote will truly influence the outcome of the election. In my sample of elections, average voter turnout in executive elections is slightly lower in more democratic countries (those on the positive spectrum of Polity) than in more autocratic countries but the difference is marginally significant—64 percent to 68 percent. For legislative elections, the difference is not significant. When voter turnout and violence are compared across democratic and autocratic countries, there is some evidence that electoral violence has a differential effect on voters in non-democratic settings. In autocratic executive elections held in Africa where no violence took place prior to Election Day, average voter turnout was 57 percent. In autocratic executive elections where pre-election violence did take place, voter turnout was, on average, 71 percent. This is suggestive of a mobilizing effect of violence for non-democratic elections, but this relationship does not hold for legislative elections. There is no substantial difference in turnout based on election violence for legislative elections in the sample. Conclusions

Having examined data on voter turnout cross-nationally and subnationally in the case of Kenya, I have yet to find a consistent, significant effect of electoral violence on voter turnout figures in the aggregate. At the individual level, however, there is evidence that electoral violence affects willingness to vote, but this effect differs by partisan attachment and specific electoral context. Although not conclusive, there is evidence to suggest that different types of electoral violence produce different types of effects on potential voters. In some cases violence mobilizes voters and in other cases it discourages them from participating in future elections. The differences across individuals and countries might be a result of how directly affected individuals are by violence, the identities of the main perpetrators of violence, or the message that violence is meant to convey to voters. Strategic and incidental electoral violence can send vastly different messages to the electorate depending on the main actors and their grievances. Strategic violence might be signaling to voters that candidates or parties

144

Electoral Violence and Voter Turnout

are willing to win at any cost. Incidental electoral violence could be signaling the growing strength of the opposition or fear of electoral loss on the part of the ruling party. This complex relationship helps to explain why it is difficult to detect the effects of electoral violence using national voter turnout figures. If we assume that politicians are rational actors, then they either want to win with as many votes as possible or with as few dissenting votes as possible. Violence, it seems, can achieve both objectives. It is important to note that while the stated willingness to vote does seem affected by electoral violence, it is impossible to determine how or if vote choice is affected. It may be the case that voters are participating and casting their ballots for candidates or parties that they do not necessarily personally endorse or would choose in the absence of violence. It may also be the case that violence changes partisan attachments. Unfortunately, this is not a proposition that can be tested with available data. Electoral violence does not always have the high profile of a civil war: rarely are elections as violent as Kenya 2007, Republic of Congo 1994, or Côte d’Ivoire 2010. But the persistence of violent elections reinforces a norm of political transition that will eventually be at variance with the ideals and goals of democracy, especially as free, fair, peaceful, and non-coercive elections are a necessary prerequisite for democratic development. In pursuit of further understanding the consequences of electoral violence, it is necessary to next examine the impact electoral violence has on how individuals assess and relate to governments and regimes because it is these relationships that form the keys to democratic consolidation. Notes

                                                                                                            1

Levitsky and Way, “The Rise of Competitive Authoritarianism.” Lindberg, “Democracy by Elections.” 3 Lijphart, “The Problem of Low and Unequal Voter Turnout.” 4 Voter Turnout Database. Figures are for turnout of registered voters. 5 Voter Turnout Database. Figures are for turnout of registered voters. 6 Bratton, “Vote Buying and Violence in Nigerian Election Campaigns,” p. 2

626.

7

Collier and. Vicente, “Votes and Violence.” Collier and. Vicente, “Votes and Violence,” p.18. 9 Travaglianti, Threatening Your Own. 10 Atubi, “Election-Related Violence in Africa;” and Meadows, “Political Violence and the Media.” 11 Hafner-Burton et al., “When Do Governments Resort to Election Violence?” 8

Electoral Violence in Sub-Saharan Africa

 

145

12 Bhasin and Gandhi, “Timing and Targeting of State Repression in Authoritarian Elections.” 13 Straus and Taylor, “Democratization and Electoral Violence in SubSaharan Africa, 1990-2008.” 14 The data come from the 47 out of 48 countries found in sub-Saharan Africa as only forty-seven have held elections during this time period. Eritrea is the only country to not hold elections. 15 “African Elections Database”; “Election Guide: Voter Turnout,”; Lindberg, Democracy and Elections in Africa; “PARLINE Database on National Parliaments”; Voter Turnout Database. 16 Pre-election violence was measured using a binary coding scheme where 1 represents districts where some form of violence related to the election occurred within the 6 month period prior to the December 27, 2007 election. All other districts were coded 0 for a non-occurrence. The temporal component of this coding scheme is consistent with the approach taken by Straus and Taylor, “Democratization.” 17 Only one figure for turnout for the presidential and legislative elections in 2007 has been released. This data analysis represents the official turnout data released by the Electoral Commission of Kenya. Additionally, it should be noted that rampant fraud is believed to have taken place which affects our confidence in the data. Nonetheless, this is the only data available. 18 “The Waki Report.” 19 Straus and Taylor, “Democratization and Electoral Violence in SubSaharan Africa, 1990-2008.” 20 These data are only available at the district level, not at the constituency level. 21 The 20 countries included in the Round 4 survey are: Benin, Botswana, Burkina Faso, Cape Verde, Ghana, Kenya, Lesotho, Liberia, Madagascar, Malawi, Mali, Mozambique, Namibia, Nigeria, Senegal, South Africa, Tanzania, Uganda, Zambia and Zimbabwe. 22 Afrobarometer Round 4. 23 International IDEA 24 “On the Brink of the Precipice.” 25 “On the Brink of the Precipice,” p. 34. 26 Cutts and Fieldhouse, “Is Voting Habit Forming?”; Smets and van Ham, “The Embarrassment of Riches?”. 27 Andre Blais,“What Affects Voter Turnout?” Annual Review of Political Science 9 (2006): 111–25. Benny Geys. “Explaining Voter Turnout: A Review of Aggregate-Level Research,” Electoral Studies 25 (2006): 637–63. 28 Borenstein et al., “A Basic Introduction to Fixed-Effect and RandomEffects Models for Meta-Analysis.” 29 See Appendix for full logistic regression results.

7 Electoral Violence and Democratic Attitudes

To be clear, the argument thus far has not been that electoral violence has no effect on individual-level behaviors; rather, that voter participation in the aggregate seems to be negligibly affected. Electoral violence does demonstrate a significant effect on voter participation, as measured through survey responses. Depending on individual partisanship, electoral violence might suppress or mobilize voter support. On its face, this finding may seem counter-intuitive. When intimidation and violence are used to coerce voters, it may be logical to think that abstention or vote prevention is the primary goal. However, elections are not won through voter abstention or disenfranchisement but many times due to efforts at voter mobilization. In addition, other individual-level factors may be affected but are difficult to detect. For example, vote choice (which party/candidate a voter supports) may be affected and compromised. The decision to register and participate at all may be precipitated by beliefs about the safety and peacefulness of the voting environment. Some researchers have insisted that the holding of elections— regardless of quality or conduct—leads to democratic gains over time.1 Successive elections are believed to lead to expansions of freedoms and liberties and a strengthening of democratic norms. Taken to its logical extreme, this argument implies that even if fraud, manipulation, intimidation, or even violence attend an election, the mere fact that an election has been held should have a generally positive effect on the democratization process. It stands to reason, however, that electoral violence— which characterizes more than half of all elections held in Africa—may exert a negative effect on the democratization process. This may be especially true when the intent of electoral violence is voter manipulation and distortion of preferences.

147  

148

Electoral Violence and Democratic Attitudes

This chapter uses public polling data of citizens across sub-Saharan Africa to examine the relationship between election violence and individual-level democratic disposition. The data show that individuals who have experienced electoral violence or fear that it will take place in the future are less likely to believe that their country is democratic. They also report lower levels of democratic satisfaction and lower levels of support for democracy in the abstract. Finally, more fearful individuals report lower levels of trust in most every salient political institution: the presidency, the legislature, the courts, the electoral management body, and the security forces. These findings suggest that violent elections, rather than advancing democracy, may actually undermine its development through the erosion of democratic satisfaction and democratic support. Democratic Development

Democracy is a system of government in which ultimate political authority rests with the citizens. In representative democracy, leaders— who are chosen by and held accountable to the citizenry—make decisions on their behalf. Specific institutional configurations vary across regimes but there are a few commonalities. Namely, decision-making powers are distributed across different actors and citizens have direct input into the selection of certain political leaders through organized elections. In addition, individual liberties and rights are explicitly protected but, again, the specific liberties and rights vary by democracy. The definition of democracy is nebulous because it is meant to encompass a variety of different regime types (parliamentary, presidential, hybrid, etc.) but all, ultimately, respect the sovereignty of the individual. Democracy is an ideal-type and, as such, democratic quality varies from country to country and within countries over time. Multiparty elections are a relatively recent phenomenon for many African countries. Prior to the late 1980s, most countries in sub-Saharan Africa were governed by some form of non-democratic rule. When elections were held, electoral competition was severely regulated. Elections were generally organized as single party affairs or as a fait accompli in which the outcome was predetermined. It was not until the end of the Cold War that the so-called “third wave of democracy” spread to Africa and the holding of multiparty elections became the rule, not the exception.2 The nearly universal embrace of elections, however, does not mean that democracy has been universally adopted as well, nor does it mean that the quality and conduct of elections is uniform across the continent.

Electoral Violence in Sub-Saharan Africa

149

Botswana, for example, has held relatively open, free and peaceful elections since 1965. Nigeria, on the other hand, has alternated between multiparty elections and military dictatorships several times over the past 55 years. Since the re-introduction of multiparty elections in 1999, most elections have been plagued by accusations of fraud, intimidation and violence including a significant number of fatalities. After the 2011 election, at least 800 were killed protesting the results.3 The 2015 elections were fraught with controversy and the specter of violence loomed so large that it made some question whether elections were more trouble than they were worth. While the outcome was ultimately accepted, there are estimates that anywhere between 50 and 200 were killed in the period before elections.4 While Nigeria’s experience with violent elections is not necessarily unique in Africa, electoral violence does not occur in every election or every country on the continent. Ghana, Senegal, and Zambia have held all very competitive elections in which a peaceful transfer of power between political adversaries has taken place. However, not every peaceful election is competitive or free. Countries such as Equatorial Guinea, The Gambia, and Rwanda routinely hold relatively peaceful elections but the opposition stands little genuine chance of dislodging the incumbent government and attaining power. These examples should underscore an important point that is often overlooked when discussing democratic development: elections, while a meaningful component of democracy, are in and of themselves not democracy. Equating the two is deeply problematic. It is because of this nuance that researchers are interested in untangling the exact nature of the relationship between elections and democracy. Specifically, much current debate is focused on understanding the causal connection between the holding of elections and democratic quality. According to some, the simple act of holding elections—regardless of quality, conduct or outcome—leads to democratic gains in terms of expansions of freedoms and liberties, and a strengthening of democratic norms over time. Holding successive elections, it is argued, creates a self-reinforcing cycle in which democratic norms are learned and habituated with each subsequent election. Democratization is then viewed as a cumulative process in which each electoral cycle, regardless of how flawed, provides electoral actors important information and experiences which they can use to improve future elections. Lindberg repeatedly tests this proposition with data from sub-Saharan Africa where he finds consistent support for the democratizing effect of elections.5 Some researchers have questioned this proposition on both empirical and theoretical grounds. Bogaards (2013), for example, systematical-

150

Electoral Violence and Democratic Attitudes

ly re-examines much of the data from sub-Saharan Africa that Lindberg (2006; 2009) used in order to demonstrate that elections, on the contrary, seem to occupy a nebulous relationship between democratic procedure and democratic quality.6 He is only able to identify three cases on the continent in which repeated elections may have led to a deepening of democracy; in the vast majority of cases, elections reinforce the nature and quality of the regime holding them. Elections held by authoritarian regimes entrench autocracy and elections held by liberalized regimes promote the expansion of various freedoms. Taken to its logical extreme the argument implies that even if fraud, manipulation, intimidation, and violence attend an election, the mere fact that an election has been held will have some positive effect on democratic maturation. Greenberg and Mattes (2013), in analyzing how the quality of elections affects democratization, find that individual assessments of the freeness and fairness of elections directly affect how individual assess the quality of democracy.7 Those who perceive their country’s elections to be fraudulent are much more likely to believe that their country is not a democracy. It stands to reason that voter intimidation, harassment, and violence—which together characterize more than half of all elections held in Africa—could also have a negative effect on democratic attitudes. Electoral violence that is employed to manipulate voters may be particularly pernicious. Deliberate or strategic electoral violence might create a negative impression on voters in terms of the specific actors believed to partake or the process itself. Significant election-related losses may convince a voter that elections are more trouble than they are worth. A voter who is repeatedly exposed to threats of violence around election time could eventually become disenfranchised or disillusioned with the democratic process entirely. In order to further the debate regarding the relationship between elections and democracy, this chapter uses polling data of citizens from 20 countries in sub-Saharan Africa to examine the relationship between election violence and attitudes toward democracy. The implications of the answer to this question to the foreign policy community are legion, especially if it is the case that supporting elections under any and all circumstances may do more harm than good in terms of democracy promotion. Rushing to hold elections in countries that are unable or unwilling to protect the integrity of the process and its participants could undermine the promotion of democracy, a goal the international community has come to explicitly prioritize.

Electoral Violence in Sub-Saharan Africa

151

Democratic Attitudes: Satisfaction and Support

Because of the emphasis on citizen consent and participation, democracy is sustained in no small part by the acceptance of the citizenry that this method of solving inevitable political conflict is the best method to do so. In comparison, autocracies rely on much different sources of leadership selection, legitimacy, and methods of regime maintenance. Military dictatorships, for example, rely on the strength and power of the security apparatus to stay in power. Personal dictatorships rely on the power of an individual leader and his or her personality/favorability to stay in power. Additionally, many autocracies rely on patronage networks or selective kinship relationships for public support. As democracies require some level of citizen consent, democratic satisfaction and support are important subjects of study. Some have argued that durable satisfaction with democracy takes time to develop. Citizens in older, more mature democracies have generally been found to be more satisfied than citizens in newer democracies.8 Newer research, however, calls this question into finding. Doorenspleet (2012) looks specifically at democratic satisfaction in 8 African countries.9 According to her, more than half (54.6 percent) of the respondents she examines are satisfied with democracy and these cases include both older democracies (Botswana) and newer democracies (Benin) in Africa. She argues that the levels of democratic satisfaction found in sub-Saharan Africa are comparable to those found in the older regimes such as those in Western Europe, where the majority of previous research on democratic satisfaction had taken place. In a sample of 11 European countries that Anderson and Guillory (1997) examined in the 1990s the mean value of democratic satisfaction is 59.2 percent, only a few percentage points higher than the mean value of Doorenspleet’s data.10 Countries that perform better in terms of economic growth, wealth distribution, and service provision also tend to enjoy higher levels of satisfaction and democratic support. Sarsfield and Echegaray (2005) examine the sources of democratic satisfaction and support in Latin America, a region that has had recent experience with non-democratic regime alternatives.11 Based on their analysis of public opinion data taken from 8 Latin American countries in 1995, support for democracy is not intrinsic, or based on a normative preference for democracy’s ideals, but rather based on what benefits it produces. However, they are clear that erosion in democratic support does not immediately translate into a preference for non-democratic alternatives. Rather, it may lead to regime apathy in the short to medium-term and problems in the long-term if democracy is, over time, unable to produce significant gains in econom-

152

Electoral Violence and Democratic Attitudes

ic or social development. They find that those who report being more satisfied with democracy are more likely to prefer it to other autocratic alternatives while that those who are dissatisfied are more likely to exhibit apathy about the type of regime that they live under. Material benefits are not the only measure of government performance. As Greenberg and Mattes (2013) argue, the quality of elections may affect how individuals assess the quality of democracy.12 Electoral quality can be considered a reflection of a government’s ability to coordinate, organize, and administer a public good. As such, it makes sense that quality of election affects democratic assessments as well. While fraud is one measure of electoral quality, another important indicator is level of harassment, intimidation, and/or violence that accompanies an electoral event. Although electoral violence may be employed primarily for the purpose of affecting electoral outcomes, it is possible that there are other, perhaps unanticipated effects produced when actors seek to intimidate and manipulate the preferences of voters. When certain regimes or political actors are believed to have engaged in violent electoral strategies, voters may come to hold negative impressions of these actors and/or elections. Over time, if elections routinely become associated with predictable patterns of violence—as is unfortunately the case in many new democracies—support for elections in general may wane making countries more vulnerable to regime change. Data Analysis

Survey data is arguably one of the best sources of information on how individuals relate to and assess the quality of democracy in their country. For the purposes of this analysis, data from Rounds 4 and 5 of the Afrobarometer are utilized. Round 4 was conducted from 2008 to 2009 in 20 countries and Round 5 from 2011 to 2013 in 35 countries. Surveys from these two rounds specifically address the issue of election violence by asking the question: “During election campaigns in this country, how much do you personally fear becoming a victim of political intimidation or violence? A lot, Somewhat, A little bit, or Not at all.” Figure 7.1 displays the range of responses to this question across the sample of 20 cases in Round 4. Fear of election violence in this case is a positive response to the question in the survey, meaning that a respondent answered “a lot,” “somewhat,” or “a little bit” to the question.

Electoral Violence in Sub-Saharan Africa

153

Figure 7.1: Fear of Electoral Violence in Africa, 2008 to 2009

Botswana Cape Verde Benin Mozambique Senegal Malawi Burkina Faso Lesotho Madagascar Mali Ghana Tanzania **Average** South Africa Liberia Zambia Namibia Nigeria Uganda Kenya Zimbabwe 0

20

40

60

80

100

Afrobarometer Round 4, 2008-2009

On average, 54 percent of African respondents reported that they feared election violence to some extent. Respondents in Botswana, one of Africa’s oldest and most mature democracies, reported the lowest levels of fear of election violence (16 percent) compared to respondents in Zimbabwe, 90 percent of whom reported fearing election violence. Although elections have been held regularly since 1980, many of Zimbabwe’s contests have been fraught with intimidation and violence; all have returned Robert Mugabe and the Zimbabwe Africa National UnionPatriotic Front (ZANU-PF) to power. The 2008 election took place shortly before Round 4 of the survey was conducted. The election was highly competitive and tensions were high throughout the country. Be-

154

Electoral Violence and Democratic Attitudes

tween the first and second rounds of voting, at least two hundred opposition supporters were murdered and hundreds of thousands were forcibly displaced from their homes.13 Figure 7. 2: Fear of Electoral Violence in Africa, 2011-2013

Cape Verde Niger Madagascar Botswana Mauritius Algeria Burkina Faso Morocco Mali Lesotho Namibia Cameroon Mozambique Benin Malawi Ghana South Africa Burundi **Average** Swaziland Liberia Tanzania Togo Nigeria Senegal Sierra Leone Uganda Zambia Guinea Cote d'Ivoire Kenya Zimbabwe

Fear Electoral Violence

0

20

40

60

Afrobarometer Round 5, 2011-2013

80

100

Electoral Violence in Sub-Saharan Africa

155

Figure 7.2 displays responses to the same question asked of respondents in Round 5 of the Afrobarometer Survey. Zimbabweans still reported the highest levels of fear of election violence (80 percent) whereas residents of Cape Verde, Niger, and Mauritius reported the lowest levels (17 percent, 18 percent and 24 percent, respectively). The average percent of all respondents stating that they feared election violence declined from the previous survey, from 54 in Round 4 percent to 47 percent in Round 5. The Afrobarometer survey also asks respondents several questions regarding the quality of democracy in their home country including an individual’s personal satisfaction with democracy. As Table 7.2 demonstrates, satisfaction with democracy varies over time and country. In Round 4, citizens from Botswana reported the highest levels of democratic satisfaction. Citizens from Madagascar, Zimbabwe, and Senegal all reported the lowest levels of democratic satisfaction. The results from Round 5 (which have only been released on a preliminary basis) again show Zimbabwe toward the lower end of the democratic satisfaction spectrum and Botswana toward the higher end of the spectrum. In Round 4, slightly less than half of respondents reported being satisfied with democracy. In Round 5, this number increased to 58 percent of respondents. While looking at the variation in democratic satisfaction and fear of election violence across these cases may be anecdotally suggestive of a relationship, statistical analysis is required in order to probe the significance and strength of this relationship. Additionally, even though fear of electoral violence may influence an individual’s democratic attitudes, or democratic disposition, there are several other factors that are believed to impact it as well. As mentioned previously, other performance-based indicators such as assessment of the national economy should exert an influence on selfreported democratic attitudes. Individual level factors such as educational attainment, age, and geographic location are also believed to affect satisfaction with democracy as they all affect how individuals perceive the government and interact with it. Finally, much recent research has found a link between electoral outcomes and democratic attitudes. Specifically, how well an individual’s preferred party or candidate performs in elections is also related to how individuals assess democratic quality.14 Variables measuring these factors can either be found in the survey itself or constructed based on the results of the most recent past election. Ordered logistic regression analysis is used to test the hypothesis that fear of election violence is negatively related to an individual’s democratic disposition. Three different measures of democratic orientation were used as a robustness check on the relationship. Taken together,

156

Electoral Violence and Democratic Attitudes

an individual’s orientation to democracy, or democratic disposition, is represented by the responses to the questions “how satisfied are you with the way democracy works in your country,” “how much of a democracy is your country today,” and if democracy is preferred over nondemocratic alternatives. Because of the categorical structure of the dependent variables, ordered logit is the most appropriate method of statistical analysis. Each model examines these dependent variables in turn. Fear of electoral violence is the key independent variable in all three models. It is an ordinal variable ranging from 1 to 4 which captures whether an individual reported feeling afraid of violence around election periods or not. While there are ways of objectively measuring the occurrence of electoral violence, using individual level responses ensures variation within countries as well as between. This method, however, does not allow for a more nuanced analysis of the differential effects of incidental and strategic violence. Controls for other performance assessments and individual characteristics are included. Assessment of past economic performance, urban or rural inhabitance, age, gender, education, and frequency one listens to radio news are all included in the three models. I chose to use random effects for these models, which allows data from all 20 countries to be analyzed together while also incorporating the assumption that some unmeasured country-specific factors may also affect individual citizens’ attitudes.15 Table 7.1 reports the results of three ordered logit regressions that test the relationship between fear of electoral violence and democratic orientation using Round 4 Afrobarometer data.16 In all three models, the performance-based variables exert a significant impact on democratic disposition. Fear of electoral violence is statistically significant and related to all three measures in the expected direction. Those who fear election violence are less likely to report feeling satisfied with their democracy, less likely to assess their country as a democracy, and less likely to prefer democracy to other regime types. Economic performance is also significant in all three models, but to a lesser extent. Those who have more favorable economic assessments are more satisfied with democracy, more supportive of democracy, and assess the quality of their democracy higher. These findings indicate that democratic attitudes and orientation are related to how governments perform their essential functions, and these functions are not limited to purely economic concerns but also include the conduct of elections.

Electoral Violence in Sub-Saharan Africa

157

Table 7.1: Fear of Electoral Violence and Democratic Disposition, Random Effects Ordered Logit Regressions (Odds Ratios Reported) Model 1: Democratic Satisfaction

Model 2: Democratic Extent

Model 3: Democratic Support

Fear of Electoral Violence

.80***

.81***

.95**

Ruling Party Supporter

1.71***

1.75***

1.07***

Fear X Ruling Party

1.06*

1.00

.94*

Opposition Party Supporter

.95

1.04

1.17**

Fear X Opposition Party Supporter

1.00

.96

1.03

Economic Assessment

1.20***

1.15***

1.04**

Urban Residence

1.19***

1.08***

1.02

Age

1.00***

.99

1.00***

Gender

1.00

1.15***

.84***

Education

.95***

.93***

1.07***

Radio News

1.05***

1.03***

1.08***

Log Likelihood

-32230.35

-28436.14

-17112.54

n

24,239

23,723

24,338

*p=.05 **p=.01 ***p=.001

Fear of election violence has a consistent and negative impact across all three measures of democratic disposition for the 20 countries analyzed. Unlike with voter turnout, partisanship appears to exert no consistent

158

Electoral Violence and Democratic Attitudes

mediating effect on the relationship between fear of electoral violence and democratic attitudes. The coefficients estimated for the interaction between partisanship and fear of electoral violence are insignificant across the three models. The attitudes of ruling party supporters, opposition party supporters, and non-partisans are all uniformly—and negatively—affected by fear. However, in general, ruling party supporters exhibit higher levels of democratic assessment, satisfaction, and support than do opposition supporters and non-partisans. Democratic satisfaction, democratic extent, and democratic support are each measured as an ordinal variable in order to capture varying levels of satisfaction, assessment, and support. As such, the coefficients on electoral violence can be teased out to demonstrate just how much of an impact fear of electoral violence has on each aspect of democratic disposition. Table 7.2 displays the predicted probabilities for the three attitudinal variables using the above regression results, with all other control variables held at their mean values. Table 7.2: Electoral Violence and Democratic Disposition in Africa, 2008-2009 (Predicted Probabilities) If Individual Does Not Fear Electoral Violence

If Individual Fears Electoral Violence

~30%

~21%

~40%

~29%

~58%

~50%

Democratic Assessment Probability They Believe Their Country Is a Democracy Democratic Satisfaction Probability They Are Satisfied with Democracy Democratic Preference Probability They Prefer Democracy Afrobarometer 4, 2008-2009

Electoral Violence in Sub-Saharan Africa

159

These predicted values show how democratic attitudes change depending on whether one fears electoral violence or not. For example, if an individual fears electoral violence, they are approximately 9 percent less likely to believe that their country is a democracy. Fear of electoral violence also decreases the probability that an individual will be satisfied with democracy by roughly 11 percent. Fear of electoral violence is related to an approximate 8 percent decrease in the probability an individual will prefer democracy to non-democratic alternatives. When Kenya, Senegal and Liberia were analyzed separately, the relationship between electoral violence and democratic attitudes remains generally the same, but the impact of electoral violence becomes much more pronounced in the case of Kenya. When analyzed on its own, the relationship between electoral violence and democratic disposition in Kenya is much stronger than in all other countries in the sample. Democratic satisfaction decreases by close to 25 percent if one reported being afraid of electoral violence in Kenya shortly after the post-election violence occurred (Table 7.3). Taken together, these results indicate that fear of electoral violence significantly affects how individuals assess and internalize the quality of democracy in their country. Table 7.3: Electoral Violence and Democratic Disposition in Kenya, 2008 to 2009 (Predicted Probabilities)

If Individual Does Not Fear Electoral Violence

If Individual Fears Electoral Violence

~28%

~17%

~81%

~56%

Democratic Assessment Probability They Believe Their Country Is a Democracy Democratic Satisfaction Probability They Are Satisfied with Democracy Afrobarometer 4, 2008-2009

160

Electoral Violence and Democratic Attitudes

Surprisingly, fear of electoral violence is no longer significantly related to support for democracy in Kenya. This may be a reflection of the international intervention that occurred after the PEV in 2008. The resultant peace agreement stipulated that a new constitution must be drafted and enacted prior to elections being held. Kenyans may have broadly interpreted this as a reset of their democracy, which may account for the null finding. The relationship between democratic attitudes in Senegal and Liberia look very similar to those estimated by the cross-national dataset in so far as the effects are much more muted than they are Kenya. This finding makes sense given the extreme nature of the electoral violence in Kenya. These results might also reflect a growing tension between support for democratic ideals or democracy in theory and democratic underperformance in this area by their respective governments. Election violence is alienating citizens from their own processes and may translate into longer term problems with legitimacy. Table 7.4: Political Trust and Fear of Electoral Violence in Africa Fear of Electoral Violence

No Fear

Trust in Presidency (a lot)

31%

45%

Trust in Legislature (a lot)

22%

31%

Trust in EMB (a lot)

20%

31%

Trust in Ruling Party (a lot)

23%

34%

Trust in Opposition Party (a lot)

14%

15%

Trust in Courts (a lot)

26%

35%

Trust in Police (a lot)

22%

32%

Afrobarometer Round 4, 2008-2009

Electoral Violence in Sub-Saharan Africa

161

Fear of election violence is also significantly and negatively related to institutional trust. This relationship holds across several different political institutions including the presidency, the legislature, the courts, election management bodies, and security services. Table 7.4 reports the bivariate relationship between fear of electoral violence and levels of trust in several different political institutions for the sample of 20 Afrobarometer countries surveyed in Round 4. With the exception of trust in opposition parties, every other reported type of institutional trust is associated with approximately a 10 percent decrease between those who fear electoral violence and those who do not. Table 7.5: Political Trust and Fear of Electoral Violence in Kenya Fear of Electoral Violence

No Fear

Trust in Presidency (a lot)

27%

45%

Trust in Legislature (a lot)

18%

27%

Trust in EMB (a lot)

9%

15%

Trust in Ruling Party (a lot)

13%

21%

Trust in Opposition Party (a lot)

8%

11%

Trust in Courts (a lot)

13%

18%

Trust in Police (a lot)

11%

13%

Afrobarometer 4, 2008-2009

The strong correlation between trust and electoral violence does not indicate the direction of causation. It is just as possible that a lack of trust in various political institutions leads to fear of election violence as it is that election violence reduces the trustworthiness of institutions. Nonetheless, the relationship leads to the same conclusion: election

162

Electoral Violence and Democratic Attitudes

violence and political institutions needs to be addressed simultaneously. A deficiency in either can negatively impact the growth of democracy. Looking to Kenya specifically (Table 7.5), where extreme electoral violence virtually paralyzed the country, this relationship is even more pronounced. Levels of trust in Kenya are generally lower than other countries. For example, amongst those who report not being afraid of elections, 45 percent state that they have a lot of trust in the presidency. Amongst those who do fear elections, however, trust in presidency is almost 20 points lower: 26.9 percent state that they have a lot of trust in the presidency. The relationship is weaker when looking at trust in political parties, courts, and the police. In Kenya, the survey was conducted less than a year after the post-election violence in which the Kenya Electoral Commission, political parties, police, and courts were all heavily criticized for their role in fomenting violence. Although not as robust as in Kenya, this relationship maintains in both Liberia and Senegal as well. This difference in magnitude of effect may be related to the different types of electoral violence that have occurred in each case (incidental or strategic) or to the intensity of violence across these three cases. Conclusions

After decades of various types of dictatorship, autocracy, and even a brief attempt at empire (in the Central Africa Republic), elections have supplanted all other methods of leadership selection and have become the norm in sub-Saharan Africa. Elections may be one of the few unchallenged policy prescriptions the international community at large has come to endorse as a method of establishing legitimate government after some form of regime interruption. In the aftermath of such disparate events as coups, civil conflict, and/or civil war, the general consensus seems to be that elections are held as soon as possible. However, recent research suggests that the rush to elections may be counterproductive. Brancati and Snyder (2012), for example, find that postconflict elections held before sufficient reforms have taken place, are more likely to result in a resumption of conflict. Electoral violence, an unfortunate feature of more than half of all elections held in Africa, has a negative effect on several different measures of democratic assessment, satisfaction, and support in the more open African countries. Even in countries with relatively lower levels of election-related violence such as Senegal and Liberia, fear of election violence has a negative effect on democratic satisfaction, assessment, and support. Unsurprisingly, the negative effects of electoral

Electoral Violence in Sub-Saharan Africa

163

violence on individual attitudes towards democracy are much stronger and more pronounced in Kenya than in Senegal or Liberia. Election violence in Kenya is typically much more intense and fatal than it has been in Senegal or Liberia. Democratic support is the only democratic attitude that does not appear to be consistently and significantly related to fear of election violence. In Kenya, specifically, one potential explanation has to do with the intervention and response that took place a result of PEV, both of which occurred before the Round 4 survey was administered. A substantial change in political institutions may have mitigated the deleterious effects of electoral violence on support for democracy. However, this relationship may not hold if elections continue to prove violent in the future. Another interpretation is that democracy as promised continues to be supported by the vast majority of citizens surveyed by Afrobarometer. But the consistent relationship between fear of electoral violence and democratic assessment and democratic satisfaction across countries and time periods indicates that citizens’ fear of the electoral process affects how they perceive and relate to their governments. Because this analysis is roughly limited to a specific period of time, I cannot adequately assess how the relationship between violence and democratic disposition evolves or changes over time or after several rounds of elections. Nonetheless, there is sufficient evidence to suggest that fear of electoral violence has a significant and negative effect on individual-level democratic attitudes in the period following elections. In the case of Kenya, this relationship was measurable in a survey that took place more than three years after significant electoral violence occurred. Due to a lack of available data, it also remains to be seen how this relationship might play out in more closed and autocratic political systems. Individuals might not associate electoral violence with democratic support or satisfaction in a non-democracy. On the other hand, because elections are often conflated with democracy, violence that occurs during non-competitive elections could prove even more of an indictment of democracy. Despite these outstanding questions, there is no doubt that the conduct of elections matters to citizens. Individuals who have experienced electoral violence or fear that it will take place are less likely to believe that their country is democratic. They also report lower levels of democratic satisfaction and, in some cases, lower levels of support for democracy in the abstract. Finally, they report significantly lower levels of institutional trust. These findings suggest that elections, when violent, rather than advancing democracy may actually undermine its develop-

164

Electoral Violence and Democratic Attitudes

ment as individuals use election violence as a reference point for relating to their own democracy. The fact that partisanship does not intercede in the relationship between fear of election violence and democratic disposition underscores just how deeply problematic electoral violence is. Whereas violence can compel or deter voters (i.e., some voters may be more inclined to vote if they fear election violence and some may be less inclined to vote if they fear election violence), democratic attitudes are uniformly adversely affected by violence. Whether one is a supporter of the ruling party, an opposition party supporter, or a self-identified non-partisan, fear of electoral violence is negatively related to democratic disposition. Regardless of one’s own partisan attachments, electoral violence exhibits a negative effect on how one perceives and relates to democracy. Therein lies the biggest danger violence poses to democracy: erosion of public trust, public confidence, and legitimacy across all sectors of society. Notes

                                                                                                            1

Lindberg, Democracy and Elections in Africa; Lindberg, “The Surprising Significance of African Elections”; Lindberg, ed., Democratization by Elections: A New Mode of Transition; Lindberg, “Democratization by Elections? A Mixed Record.” 2 Huntington, The Third Wave. 3 “Nigeria: Post-Election Violence Killed 800.” 4 “A Pre-Election Report and Advisory on Violence in Nigeria’s 2015 General Elections.” 5 Lindberg, Democracy and Elections in Africa; Lindberg, “The Surprising Significance of African Elections”; Lindberg, ed., Democratization by Elections: A New Mode of Transition; Lindberg, “Democratization by Elections? A Mixed Record.” 6 Boggards, "Reexamining African Elections." 7 Greenberg and Mattes, “Does the Quality of Elections Affect the Consolidation of Democracy?” 8 Anderson and Guillory, “Political Institutions and Satisfaction with Democracy”; Anderson, “Parties, Party Systems, and Satisfaction with   Democratic Performance in the New Europe”; Lagos, “World Opinion: Support for and Satisfaction with Democracy.” 9 Doorenspleet, “Critical Citizens, Democratic Support and Satisfaction in African Democracies.” 10 Anderson and Guillory,“Political Institutions and Satisfaction with Democracy,” p. 70. 11 Sarsfield and Echegaray, “Opening the Black Box: How Satisfaction with Democracy and Its Perceived Efficacy Affect Regime Preference in Latin America.”

Electoral Violence in Sub-Saharan Africa

 

165

12 Greenberg and Mattes, “Does the Quality of Elections Affect the Consolidation of Democracy?” 13 “Bullets for Each of You: State-Sponsored Violence since Zimbabwe’s March 29 Elections.” 14 Blais and Gelineau, “Winning, Losing and Satisfaction with Democracy”; Singh et al., “Differentiating Winners: How Elections Affect Satisfaction with Democracy”; Howell and Justwan, “Nail-biters and No-contests: The Effects of Electoral Margins on Satisfaction with Democracy in Winners and Losers.“ 15 Borenstein et al., “A Basic Introduction to Fixed-Effect and RandomEffects Models for Meta-Analysis.” 16 Benin, Botswana, Burkina Faso, Cape Verde, Ghana, Kenya, Lesotho, Liberia, Madagascar, Malawi, Mali, Mozambique, Namibia, Nigeria, Senegal, South Africa, Tanzania, Uganda, Zambia, and Zimbabwe.

8 Electoral Violence and the Future of Democracy in Africa

Elections, in theory, are meant to provide citizens with ultimate authority over their governments by allowing them to decide who best represents their interests and whom they most trust to make decisions on their behalf. Prior to 1990, virtually all African countries were governed by some form of autocracy—military rule, single party regimes, personal dictatorships and the like. The commonality being that citizens had little input into how or who ruled them, political power was centralized and what was termed “big man” politics was the norm. Now, elections are the rule, not the exception. Many believed that the advent of multiparty elections in the 1990s signaled a new era of governance in Africa; one in which the citizen would reign supreme. Even though elections are not democracy in and of themselves, they are an essential feature and one that is frequently used as a proxy measure of democracy. Unfortunately, as this book has detailed, violence has become a frequent and persistent feature of many elections that, at its worst, could threaten to trigger war and, in a best case scenario, alienates voters and undermines citizen satisfaction and trust in democracy. This book has detailed the nature of electoral violence in three vastly different countries: Kenya, Liberia, and Senegal. All three countries have long electoral histories—Liberia and Senegal in particular have held elections since the 1800s—and yet these countries’ experiences have not translated into the types of democratic gains we might hope to see. In 1997, Liberia held multiparty elections in the midst of its civil war in order to end the conflict. It was widely hoped that successful elections in Liberia would lead to two desired outcomes: an elected government that reflected the will of the citizens and that was viewed as 167  

168

Electoral Violence and the Future of Democracy in Africa

legitimate by the country’s numerous former combatants.1 Unfortunately, neither was achieved; civil war resumed in 1999 and lasted until 2003. Two subsequent elections have been held in Liberia and the peace has endured. While the potential for violent elections has been present, it did not occur to a significant level in either the 2005 or 2011 elections. In addition to the grievances that the opposition had for engaging in election, Liberia meets most of the criteria outlined above that would suggest its elections would be violent in the period prior to Election Day. Liberia holds concurrent presidential and legislative elections. Liberia also uses majoritarian electoral rules for its presidential election and plurality rules for its legislative elections. Its democracy is new and its institutions are weak. But there is a key difference that helped to ensure the 2005 and 2011 elections were relatively calm. The commitment of political leadership and the heavy presence of the international community before, during, and after the elections made election violence untenable. As a strategy, this violence would not have achieved much, and its organizers would surely have been caught. But how long can the international community administer Liberia’s elections? How long can they provide electoral security, either directly or indirectly, through the imposing presence of a peacekeeping operation? Going forward, as the international community turns its focus to other new democracies that need its assistance, the potential for violence in Liberia is very real if steps are not taken to strengthen the capacity of domestic electoral institutions. Kenya, on the other hand, also meets most of the requirements for pre-election violence to take place (plurality rules, concurrent elections, weak institutions). And it also has an extensive history of pre-election violence and post-election violence. Furthermore, political actors who organize violence generally do so with impunity. Very few perpetrators have been convicted of electoral violence offenses. The 2013 elections, while still exhibiting significant pre-election violence, were perhaps less violent due to the increased media attention, both domestic and international, that accompanied its first postconflict election. In this respect it is similar to Liberia’s 2005 election. Senegal employs majoritarian rules for its presidential elections, which are usually held separately from legislative elections. For its legislative elections, a parallel or mixed system of apportionment was adopted in the early 1980s. Most of its elections have experienced some form of pre-election violence but it differs in kind and magnitude from that generally found in Kenya. The country’s long evolution toward transparent elections management was nearly derailed by the lack of

Electoral Violence in Sub-Saharan Africa

169

commitment of political leadership to peaceful and fair elections. Specifically, Abdoulaye Wade’s maneuvering to undermined the institutions and force himself into office for a third term led to incidental violence which only abated when political institutions forced Wade to concede defeat. The management of elections has improved markedly over time in Senegal but Wade’s ability to change the constitution on a whim and influence the electoral dispute resolution process suggests that these are two areas that need attention before the next elections take place. The 2013 election in Kenya, roundly deemed a “success,” was directly related to approximately 300 fatalities before the election and upwards of 13 on the day of the election. There were also some serious deficiencies in administration and management that call into question the veracity of the outcome. The 2012 election in Senegal was arguably the country’s most violent to date and, although the outcome was ultimately peaceful, there was a brief period where many were seriously concerned that violence could escalate and spiral out of control. Liberia, a country that is still recovering from a nasty conflict in the 1990s, has held two “peaceful” elections in the postconflict period but there is no guarantee that this trend will continue. While there are many lessons that can be gleaned from these countries experiences, perhaps the most important is that the political environment in which elections are held matters for democracy. Under plurality electoral rules—which reinforce winner take all mentalities— strategic electoral violence is more likely to take place in both executive and legislative elections. Elections held by more autocratic regimes— those who restrict competition and participation in elections and often have poor track records with respect to protection of civil liberties and individual freedoms—are much more likely to result in violence and intimidation than elections held in less repressive contexts. Poorly managed elections are much more likely to experience violence than elections that are generally perceived to be free and fair. Imbalances and partiality in administration, management, and dispute adjudication increase the likelihood that routine electoral conflicts will escalate. Independent and competent electoral institutions can dramatically reduce the likelihood that violence will occur. Incidental electoral violence is a product of electoral fraud, malfeasance, and a belief in the unfairness of the process. Improved electoral management through the strengthening of election management bodies and domestic observation help to alleviate some of the procedural concerns. But often times political elites—especially those in power—take advantage of incomplete or inadequate reforms to undermine the integrity of the vote. Election reform requires diligence to implement and take

170

Electoral Violence and the Future of Democracy in Africa

effect. In the case of Liberian elections, the international community has sometimes taken up key functions of election administration (the tallying of votes, the provision impartial electoral security) that reduce opportunities for fraud and violence. This is not, however, a long-term sustainable strategy. If domestic institutions are compromised or otherwise unable to take up these functions in the future, fraud and violence are likely to occur. Strategic election violence takes place under circumstances where the supporting electoral architecture is also poorly functioning. Captured electoral bodies, partisan security forces, inadequate judicial systems all contribute to the likelihood that strategic violence takes place. What the case of Kenya makes abundantly clear is that elections require a significant number of ancillary institutions are required to prevent violence from becoming a routine part of election campaigns. Post-election violence may be deterred by a significant presence of international actors, as was the case in Kenya’s 2013 election, but strategic violence before the election requires a different set of interventions to curtail. As successful as the 2013 elections were, approximately 300 Kenyans still lost their lives in the pre-electoral period. Despite such optimism as espoused by those who predicted the “democratizing” effect of repeatedly holding elections, democratic development on the continent seems to have stagnated2. Perhaps reflecting a larger global trend, while there has been an increase in political rights in Africa—namely in the realm of electoral participation—there has also been a simultaneous decrease in civil liberties as leaders turn to other sectors in which they can exert influence and control.3 This has become even more pronounced over the past few years as contraction of freedoms have been noted in countries such as Burundi, Tanzania, Rwanda, South Africa, Uganda, Zambia, and Ethiopia.4 As much as we might like to believe that the holding of elections can lead to democratic gains over time, there is little evidence that this is currently taking place. There have been a few bright spots to be certain, but the trend, especially recently, has been a reversion to autocracy through the ballot box. Presidents with autocratic tendencies are learning how to manipulate outcomes to their advantage and to consolidate power. Multiparty elections in Zimbabwe have returned Robert Mugabe to office five times. He has intimated that he plans to run for a sixth time in 2018. Yoweri Museveni, in power since 1986, has been elected president four times thus far and will be running again in 2016. As of early 2015, presidents in Burundi, Democratic Republic of Congo, and Rwanda all appear to be laying the groundwork to either remove or circumvent term limits so that they may remain in office.

Electoral Violence in Sub-Saharan Africa

171

In Burkina Faso, Blaise Compaoré, who had come to power in 1987 as the result of a coup, initiated a democratic transition in 1990 that led to the adoption of a new constitution in 1991. The county held multiparty elections shortly thereafter in 1992—the country’s first since 1978. Compaoré was re-elected president in 1998. And 2005. And 2010. In 2014, however, his attempts at adjusting the constitution to allow him to run for president yet again in 2015 led to a popular revolt that removed him from office. The Burkinabe experience may serve as a warning to other executives, or it may encourage others to find new, less outwardly controversial methods of clinging to power. If elections are not producing meaningful personnel changes, then what is their purpose? It is sometimes argued that elections held by autocratic regimes serve different functions than those held by democratic regimes.5 Rather than providing citizens with a mechanism to hold governments accountable, autocratic elections allow political parties to engage in patronage politics and redistribution of state resources to political supporters on a regular and predictable basis. They allow for a limited amount of competition at only the most elite of levels. They also serve to legitimize governments to both domestic and international audiences. What they do not do, however, is cause democracy. Successful and peaceful elections may require mature institutions— such as competent and impartial electoral management bodies and a vigorous independent media—that are inherently lacking in new democracies. According to Collier (2009), newly democratizing countries— especially those that are less economically developed—lack the necessary preconditions for the holding of successful and peaceful elections.6 Collier argues that democracy promotes accountability and legitimacy; however, elections by themselves do not foster accountability or legitimacy and can actually undermine both when held in environments lacking substantive protections for civil and political rights and those with weak political institutions. It is a confluence of factors that are, unfortunately, found frequently in sub-Saharan Africa. It may be that we have the sequencing of events wrong and that specific political institutions must be solidly in place before elections are adopted, especially in postconflict settings. It may also be that elections held without proper monitoring and dispute resolution mechanisms encourage violence and conflict. Finally, it may be that certain types of elections and institutional arrangements, namely elections that encourage a winner take all approach to politics, are inimical to democratic development in subSaharan Africa. Recently, there have been two edited volumes, Elections in Dangerous Places: Democracy and the Paradoxes of Peacebuilding and Voting

172

Electoral Violence and the Future of Democracy in Africa

in Fear: Electoral Violence in Sub-Saharan Africa, which have examined electoral violence in a variety of difficult cases including Afghanistan, Iraq, Nigeria, and Sudan. Both volumes underscore how hastily assembled elections have the potential to undermine the search for lasting peace in conflict-prone societies.7 Nonetheless, there is still a deep and abiding belief by the international community that elections are a necessary pre-condition for state building. For example, after a military coup in Mali deposed a democratically elected president who was, by all accounts on his way out of office, and a separatist insurrection in the North of the country led to French military intervention, the international community still pressured the country to hold elections the following summer—despite the massive displacement of hundreds of thousands of Malians and a lack of a peace agreement with rebels.8 In December 2013, as the Central African Republic (CAR) struggled to contain violence that the UN warned had some of the marking of genocidal acts after a coup took place earlier that year, France issued a statement that CAR should hold elections as soon as possible, and preferably by 2015.9 Elections in CAR are tentatively scheduled to be held in the latter part of 2015 but it remains to be seen if this will be possible. It has been argued that electoral violence is a normal part of the democratization process, that it is to be expected and not necessarily a cause for concern. This approach to electoral violence overlooks how individuals interpret and are affected by electoral violence. In some instances, incidental electoral violence occurs as an emotive response to perceived unfairness in the electoral process, signaling when and where democratic processes may be breaking down. Strategic violence, on the other hand, can be employed strategically to either prevent individuals from voting or to compel them to vote for a specific party or candidate. Ultimately, the purpose of strategic violence is to reduce competitiveness of elections and ensure electoral victory by any means necessary. The contradictory methods employed to achieve this (mobilizing and suppressive violence) mean that the effects of electoral violence on voter turnout are difficult to detect in the aggregate, but this does not mean that they do not exist. Effects manifest at the individual level in terms of willingness to vote in future elections, democratic satisfaction and support for democracy, and generalized political trust. Save the rare example, electoral violence does not usually lead to civil war or immediate regime change. But it is a persistent problem nonetheless. Other forms of political violence are roundly criticized and lead to significant consequences. Coups, for example, lead to a cessation of diplomatic relations to many Western countries. The African Union, in addition, denies membership to those countries ruled by non-

Electoral Violence in Sub-Saharan Africa

173

constitutional governments. Electoral violence, in the extreme, leads to increased attention and intervention only when it threatens to topple a government. Less intense forms of electoral violence, however, are seemingly accepted as a part of the democratization process. But this approach assumes that democracy will flourish if only elections, no matter how flawed, how violent, or how illegitimate, are adopted. And it ignores the very real effects that violence has on voters and non-voters. Electoral violence may not adversely affect specific individual’s participation in elections—it may in fact compel it—but it absolutely affects how individuals relate to democracy. Citizens who experience electoral violence report much lower levels of democratic satisfaction and support for democracy. They also report much lower levels of trust in their politicians and political institutions. Electoral violence has serious and deleterious effects on how individuals relate to democracy and their governments. Electoral violence, by attacking democracy’s primary instruments— citizens—eats away at democracy from the inside out. When citizens no longer believe in democracy, regimes become vulnerable to any manner of social and political ills. Elections are a necessary part of democratic development, to be sure, but ensuring the integrity of the electoral process is just as important. Elections held without any of the requisite protections to safeguard the integrity of the vote and the voter are mere political theater and it remains to be seen how long the audience will stick around.

 

   

Notes

                                                                                                            1

Lyons, Voting for Peace. Lindberg, Democracy and Elections in Africa. 3 Puddington, “The Democratic Leadership Gap.” 4 “Freedom in the World, 2014.” 5 Magaloni and Kricheli. “Political Order and One-Party Rule.” 6 Collier, War, Guns, and Votes. 7 Gillies, Elections in Dangerous Places: Democracy and the Paradoxes of Peacebuilding; Bekoe, Voting in Fear. 8 Nossiter, “Mali Holds Elections after Year of Turmoil.” 9 “Hollande Calls for Swift Elections in CAR.” 2

Appendix: Fear of Electoral Violence and Willingness to Vote

Fear of EV

Ruling Party Supporter

Will Vote

Will Vote

Will Vote

(Kenya)

(Senegal)

(Liberia)

-.19*

.07

-.34***

(.10)

(.09)

(.09)

.27

2.44***

2.01***

(.56)

(.31)

(.42)

Fear X Ruling Party

.61**

-.16

.14

Supporter

(.20)

(.14)

(.20)

Opposition Party Sup-

.99**

1.6***

2.4***

porter

(.45)

(.32)

(.40)

Fear X Opposition Party

.37**

.17

-.14

Supporter

(.15)

(.15)

(.16)

Voted Previously

Urban Residence

.09

-.22

.16

(.20)

(.17)

(.17)

.15

.03

-.50***

(.19)

(.15)

(.15)

175  

176

Appendix

  Age

Gender

Education

Constant

Log Likelihood

n

-.02***

-.01**

-.02***

(.006)

(.005)

(.006)

.14

-.22

-.32**

(.16)

(.14)

(.15)

.01

.12***

-.03

(.04)

(.04)

(.03)

1.12

-.59

1.88

-516.57

-666.75

-625.90

1086

1162

1163

Bibliography

Adejumobi, Said. “Elections in Africa: a Fading Shadow of Democracy?” International Political Science Review 21 (2000): 59-73. “Afrcan Elections Database.” March 2015. http://africanelections.tripod.com. Ajulu, Rok. “Kenya’s Democratic Experiment: The 1997 Elections.” Review of African Political Economy 76 (1998): 275-288. Alabi, Mojeed Olujinmi A. “The Legislatures in Africa: A Trajectory of Weakness,” African Journal of Political Science and International Relations 3, No 5 (2009): 233-241. Alexander, Harriet. “Youssou N’Dour Interview: Senegal Is Turning into a Dictatorship,” The Telegraph February 19, 2012. Accessed February 2014. http://www.telegraph.co.uk. Aldrich, John. Why Parties?: The Origins and Transformation of Political Parties in America. Chicago: University of Chicago Press, 1995. Anderson, Christopher J., “Parties, Party Systems, and Satisfaction with  Democratic Performance in the New Europe, “Political Studies 46 (1998): pp. 572-588. Anderson, Christopher J. and Christine A. Guillory. “Political Institutions and Satisfaction with Democracy: A Cross-National Analysis of Consensus and Majoritarian Systems,” American Political Science Review 91, No. 1 (March 1997): pp. 66-81. Anderson, David M. “Briefing: Kenya’s Elections 2002—The Dawning of a New Era?” African Affairs 102 (2003): 331-342. Atubi, Samuel Mondays. “Election-Related Violence in Africa,” Conflict Trends 1 (2008): 10-16. Babo, Awadh. “Tana River Violence Linked to Next Elections,” The Standard. Accessed October 2012. http://www.standardmedia.co.ke Barkan, Joel. “Kenya: Lessons from a Flawed Election.” Journal of Democracy 4 (1993): 85-99.     Barkan, Joel D. and Njuguma Ng’ethe. “Kenya Tries Again.” Journal of Democracy 9, no. 2 (1998): 32–48. Barkan, Joel D. “Kenya after Moi.” Foreign Affairs 83 (2004): 87-100. Bekoe, Dorina A. “Toward a Theory of Peace Agreement Implementation: The Case of Liberia,” Journal of Asian and African Studies 38, 2-3 (2003): 256294. Bekoe, Dorina A. (ed) Voting in Fear: Electoral Violence in Sub-Saharan Africa. Washington, DC: USIP Press, 2011. Berger, Sebastien. “Kenya’s Poll Chief Does Not Know If Kibaki Won,” The Telegraph. Accessed February 2015. http://www.telegraph.co.uk. Bergner, Daniel. “An Uncompromising Woman,” The New York Times October 22, 2010. Accessed March 2013. http://www.nytimes.com.

177  

178

Bibliography

  Bhasin, Tavishi and Jennifer Gandhi. “Timing and Targeting of State Repression in Authoritarian Elections,” Electoral Studies 32 (2013): 620-631. Binda, Stephen. “‘I Need No Degree to Perform’: Blamo Nelson Hits Back at Critics,” The Daily Observer January 25, 2012. Accessed July 2012. http://www.liberianobserver.com. Blais, André. “What Affects Voter Turnout?” Annual Review of Political Science 9 (2006): 111-125. Blais, André and Francois Gelineau, “Winning, Losing and Satisfaction with Democracy,” Political Studies 55 (2007): 425-441. “Blasts Rock Nairobi Neighborhood,” Al Jazeera December 16 2012. Accessed December 2012. http://www.aljazeera.com. Bøås, Morten. “Liberia—the Hellbound Heart? Regime Breakdown and the Deconstruction of Society,” Alternatives 22 (1997): 353-379. Boggards, Matttijs. "Reexamining African Elections," Journal of Democracy 24, No 4 (2013): 151-160. Boone, Catherine. 2011. “Politically Allocated Land Rights and the Geography of Electoral Violence: The Case of Kenya in the 1990s,” Comparative Political Studies 44, 10 (2011): 1311-1342. Boone, Catherine. “Land Conflict and Distributive Politics in Kenya.” African Studies Review 55, no. 1 (2012): 75-103. Borenstein, Michael, Hedges, Larry V., Higgins, Julian P. T., and Hannah R. Rothstein. “A Basic Introduction to Fixed-Effect and Random Effects Models for Meta-Analysis,” Research Synthesis Methods 1, Vol 2 (2010): 97-111. Bowman, Warigia and Brian Munyao Longwe. “Technology, Transparency, and the Kenyan General Election of 2013,” Al Jazeera March 29, 2013. Accessed February 2014. http://www.aljazeera.com. Michael Bratton. “Vote Buying and Violence in Nigerian Election Campaigns,” Electoral Studies 27 (2008): 621-632. 626. Bratton, Michael, Bhavnani, Ravi, and Tse-Hsin Chen. “Voting Intentions in Africa: Ethnic, Economic, or Partisan?” Commonwealth and Comparative Politics 50, No. 1 (2012): 27-52. Bratton, Michael and Mwangi S. Kimenyi. “Voting in Kenya: Putting Ethnicity in Perspective,” Afrobarometer Working Paper No. 95 (2008). Accessed March 2015. http://www.afrobarometer.org. Bratton, Michael and Eldred Masunungure. “Zimbabwe’s Long Agony,” Journal of Democracy 19, no. 4 (October 2008): 41-55. “Briefing: Kenya’s Coastal Separatists—Menace or Martyrs?” IRIN October 24, 2012. Accessed January 2014. http://www.irinnews.org. “Bullets for Each of You: State-Sponsored Violence Since Zimbabwe’s March 29 Elections,” Human Rights Watch (2008). Accessed March 2015. http://www.hrw.org. Burchard, Stephanie. “The Resilient Voter? An Exploration of the Effects of Post-Election Violence in Kenya’s Internally Displaced Person Camps,” Journal of Refugee Studies (2015). Chafer, Tony. “France and Senegal” the End of the Affair?” SAIS Review 23, No 2 (2003): 155-167. Cho, Wonbin and Michael Bratton. “Electoral Institutions, Partisan Status, and Political Support in Lesotho,” Electoral Studies 25, No. 4 (2006): 731-750.

Electoral Violence in Sub-Saharan Africa

179

Collier, Paul. Wars, Guns and Votes: Democracy in Dangerous Places. New York, NY: Harper Collins, 2009. Collier, Paul and Pedro C. Vicente, “Votes and Violence: Evidence from a Field Experiment in Nigeria,” The Economic Journal 124, No. 574 (2014): F327F355. Cooper, Helene. “On Day of Reckoning, Recalling Horror that Swallowed Liberia,” The New York Times. Accessed July 2012. http://www.nytimes.com. Cox, Gary. Making Votes Count. Cambridge: Cambridge University Press, 1997. Creevey, Lucy, Ngomo, Paul, and Richard Vengroff. “Party Politics and Different Paths to Democratic Transitions: A Comparison of Benin and Senegal,” Party Politics 11, No. 4 (2005): 471-493. Daxecker, Ursula E. “The Cost of Exposing Cheating: International Election Monitoring, Fraud, and Post-election Violence in Africa,” Journal of Peace Research 49, 4 (2012): 503-516. Daxecker, Ursula E. “All Quiet on Election Day? International Election Observation and Incentives for Pre-Election Violence in African Elections,” Electoral Studies 34, No. 10 (2014): 232-243. Daygbor, Nathaniel and Edwin Genowaye. “Liberia: CDC Bloody Clash – Doe Others Flogged,” The New Dawn August 15, 2011. Accessed March 2015. http://allafrica.com/stories/201108161697.html Derçon, Stefan and Roxana Gutierrez-Romero. “Triggers and Characteristics of the 2007 Kenyan Electoral Violence,” World Development 40, 4 (2012): 731-744. Djau, Umaro. “3 Killed in Senegal Protests, Opposition Says,” CNN. February 20, 2012. Accessed February 2013. http://www.cnn.com. Doorenspleet, Renske, “Critical Citizens, Democratic Support, and Satisfaction in African Democracies,” International Political Science Review 33 no.3 (April 2012): 279-300. Dunning, Thad. “Fighting and Voting: Violent Conflict and Electoral Politics,” Journal of Conflict Resolution 55, 3 (2011): 327-339. Duverger, Maurice. Les Partis Politiques. 2nd edition. Paris: A. Colin, 1951. “Election Guide: Voter Turnout,” International Foundation for Electoral Systems. Accessed April 2013. http://www.electionguide.org/reports1.php. Elmwood, Dennis. “Liberia: No Justice, No Peace—CDC’s Weah Wants TRC Report Implemented,” The Analyst July 12, 2012. Accessed July 2012. http://allafrica.com/stories/201207120669.html Fall, Ismaila Madior, Hounkpe, Mathias, Jindau, Adele L., and Pascal Kambale. 2011. “Senegal: Election Management Bodies in West Africa,” Open Society Foundations (2011). March 2013.http://www.afrimap.org. Ferree, Karen E.. Gibson, Clark C., and James D. Long. “Voting Behavior and Electoral Irregularities in Kenya’s 2013 Election,” Journal of Eastern African Studies 8, No.1 (2014): 153- 172. Fischer, Jeff. “Electoral Conflict and Voting: A Strategy for Study and Prevention,” IFES White Paper (2002). April 2014. http://unpan1.un.org/intradoc/ groups/public/documents/untc/unpan019255.pdf. Fjelde, Hanne and Kristine Höglund. “Electoral Institutions and Electoral Violence in Sub-Saharan Africa,” British Journal of Political Science (2014). Accessed February 2015. doi: 10.1017/S0007123414000179.

180

Bibliography

  Foeken, Dick and Ton Dietz. “Of Ethnicity Manipulation and Observation: the 1992 and 1997 Elections in Kenya.” Election Observation and Democratization in Africa, eds Jon Abbink and Gerti Hesseling. London: Macmillan, 2002. Ford, Tasmin. “Sirleaf’s Nobel Award Fails to Improve Her Street Credibility,” The Guardian UK 9 October 9, 2011. Accessed July 2012. http://www.guardian.co.uk/world/2011/oct/09/sirleaf-nobel-monrovia-row “Foreign Agents Link in Mboya Assassination.” The Standard April 15, 2012. March 2015. http://www.standardmedia.co.ke. “Former Liberian NEC Communications Chief Denies Being Opposition ‘Agent’” VOA October 31, 2011. Accessed July 2012. www.voanews.com. Frazer, Jendai and E. Gyimah-Boadi. Preventing Electoral Violence in Africa. Pittsburgh: Carnegie-Mellon University Press (2011). “Freedom in the World, 2014,” Freedom House. Accessed March 2015. https://freedomhouse.org/ Galvan, Dennis C. “Political Turnover and Social Change in Senegal.” Journal of Democracy 12, 3 (July 2001): 51-62. Galvan, Dennis C. “The Presidential and Parliamentary Elections in Senegal, February and June 2007,” Electoral Studies 28, No. 1 (2008): 492-496. Genoway, Edwin G. “Liberia: Campaign Gets Deadly,” The New Dawn August 18, 2011. Accessed March 2015. http://allafrica.com. Gettleman, Jeffrey. “Disputed Vote Plunges Kenya into Bloodshed,” The New York Times, December 31, 2007. January 2014. www.nytimes.com. Gettleman, Jeffrey. “Voting Irregularities in Kenya Election Are Confirmed, Adding Fuel to Dispute,” The New York Times March 29, 2013. Accessed March 2015. http://www.nytimes.com. Geys, Benny. “’Rational’ Theories of Voter Turnout: A Review,” Political Studies Review 4, No. 1 (2006): 16-35. Gillies, David, Elections in Dangerous Places: Democracy and the Paradoxes of Peacebuilding. Montreal: McGill-Queens’ University Press, 2011. Goldsmith, Arthur. “Electoral Violence in Africa Revisited,” Terrorism and Political Violence (2014). Accessed February 2015. Doi: 10.1080/09546553.2013.863184. Grange, Anne Kappes. “Senegal: Presidential Candidate Assesses Political Situation, Electoral Chances,” Jeune Afrique December 25, 2011: 48-51. Greenberg, Alan and Robert Mattes. “Does the Quality of Elections Affect the Consolidation of Democracy?” in Voting and Democratic Citizenship in Africa, Michael Bratton, ed. Boulder: Lynne Rienner Publishers, 2013. Haeneit-Sievers, Axel and Ralph-Michael Peters. “Kenya’s 2007 General Election and its Aftershock.” Afrika Spektrum 43 (2008): 133-144. Hafner-Burton, Emilie M., Hyde, Susan D., and Ryan S. Jablonski. “When Do Governments Use Election Violence to Stay in Power?” British Journal of Political Science 44, No. 1 (2014): 149-179. Harris, David. “Liberia 2005: an Unusual Postconflict Election,” Journal of Modern African Studies 44, No. 3 (2006): 375-395. Harris, David. Civil War and Democracy in West Africa: Conflict Resolution, Elections and Justice in Sierra Leone and Liberia. London: IB Tauris, 2011.

Electoral Violence in Sub-Saharan Africa

181

“High Stakes: Political Violence and the 2013 Elections in Kenya,” Human Rights Watch (2013). Accessed December 2013. http://www.hrw.org/sites/default/files/reports/kenya0213webwcover.pdf. Höglund, Kristine. “Electoral Violence in Conflict-Ridden Societies: Concepts, Causes, and Consequences.” Journal of Terrorism and Political Violence 21, 3 (2009): 412-427. “Hollande Calls for Swift Elections in CAR,” France24 December 9, 2013. Accessed April 2014. http://www.france24.com. Horowitz, David. A Democratic South Africa? Constitutional Engineering in a Divided Society. Berkeley, CA: University of California Press, 1991. Howden, Daniel. “Warlord to Kingmaker: An Audience with Prince Johnson,” The Independent UK October 15, 2011. June 2012. www.independent.co.uk Howell, Patrick and Florian Justwan, “Nail-biters and No-contests: The Effects of Electoral Margins on Satisfaction with Democracy in Winners and Losers,” Electoral Studies 32 (2013): 334-343. “History.” Elections Observation Group November 11, 2013. Accessed February 2014. http://www.elog.or.ke. Huntington, Samuel P. The Third Wave: Democratization in the Late Twentieth Century. Norman: OK: University of Oklahoma Publishers, 1991. Hyde, Susan D. “Catch Us If You Can: Elections Monitoring and International Norm Diffusion,” American Journal of Political Science 55, No. 2 (2011): 356-369. Hyden, Goran and Colin Leys. “Elections and Politics in Single-Party Systems: the Case of Kenya and Tanzania,” British Journal of Political Science 2 (October 1972): 389-420. “I Can Run the Country Using ICT while at The Hague, Says Ruto,” The Star October 25, 2012. Accessed February 2015. http://www.thestar.co.ke. “International Justice Monitor: Background,” Open Society Justice Initiative. Accessed March 2015. http://www.ijmonitor.org/kenya-cases-background/ “Internet World Stats.” Accessed July 2012. www.internetworldstats.com. Johnson, Martha C. “Will Senegal’s 2012 Presidential Election End in Violence?” e-International Relations February 15, 2012. Accessed January 2013. http://www.e-ir.info. Kagwanja, Peter Mwangi. “Facing Mount Kenya or Facing Mecca? The Mungiki, Ethnic Violence and the Politics of the Moi Succession in Kenya, 1987-2002,” African Affairs 102 (2003): 25-49. Kane, Ousmane Oumar. “Presidential Elections in Senegal: A Better Understanding of Recent Events,” Al Jazeera Centre for Studies April 5, 2012. Accessed March 2013. http://studies.aljazeera.net. Kante, Babacar. “Senegal’s Empty Elections,” Journal of Democracy 5, No. 1 (1994): 96-108. Karp, Jeffrey A. and Susan A. Banducci. “Political Efficacy and Participation in Twenty Seven Democracies: How Electoral Systems Shape Political Behavior,” British Journal of Political Science 38, No. 2 (2008): 311-334. Keating, Joshua E. “Foreign Policy: Is Liberia’s Future in Her Hands?” NPR August 22, 2011. Accessed July 2012. http://www.npr.org. Kelly, Catherine L. “Senegal: What Will Turnover Bring?” Journal of Democracy 23, No. 3 (2012): 121-131.

182

Bibliography

  Kelley, Judith G. Monitoring Democracy: When International Election Observation Works, and Why It Often Fails. Princeton: Princeton University Press, 2012. “Kenya: Clashes, Elections and Land—Church Keeps Watch in Molo,” IRIN July 19, 2007. Accessed March 2015. http://www.irinnews.org. “Kenya Court Upholds Uhuru Kenyatta’s Poll Win,” Al Jazeera March 31, 2013. Accessed February 2014. http://www.aljazeera.com. “Kenya: the December 29, 1992 Elections,” International Republican Institute (1993). Accessed March 2015. http://www.iri.org. “Kenya: Final Report General Elections December 27, 2007,” European Union Election Observation Mission (April 3, 2008). Accessed February 2015. http://eeas.europa.eu. “Kenya Post-Election Political Violence.” Article 19: Global Campaign for Freedom of Expression (December 1998). Accessed March 2015. http://www.article19.org. “Kenya President Mwai Kibaki Rejects MPs’ Bonus Attempt,” BBC News October 9, 2012. Accessed November 2012. http://www.bbc.co.uk/ “Kenya: Spreading the Word of Hate,” IRIN, January 22, 2008. Accessed February 2015. http://www.irinnews.org. “Kenyan Elections Marred by Mombasa Violence,” The Guardian March 4, 2013 Accessed February 2014. http://www.theguardian.com. “Kenyans Reject New Constitution,” BBC News November 22, 2005. Accessed February 2014. http://news.bbc.co.uk. King, Gary, Tomz, Michael, and Jason Wittenberg. “Making the Most of Statistical Analyses: Improving Interpretation and Presentation,” American Journal of Political Science 44 (2000): 341–355. Kramon, Eric and Daniel N. Posner. “Kenya’s New Constitution,” Journal of Democracy 22, No. 2 (2011): 89-103. Laakso, Liisa. “Insights into Electoral Violence in Africa,” in Votes, Money and Violence: Political Parties and Elections in Sub-Saharan Africa, eds. Matthias Basedau, Gero Erdmann, and Andreas Mehler. South Africa: University of Kwazulu-Natal Press, 2007. Labous, Jane. “Senegal Braces for Violence after Presidential Election,” LA Times, February 26, 2012. Accessed February 2013. http://articles.latimes. com. Labous, Jane. “Senegalese Decry Pre-Election Violence,” VOA January 28, 2012. Accessed February 2013. http://www.voanews.com. Lagos, Marta, “World Opinion: Support for and Satisfaction with Democracy,” International Journal of Public Opinion Research 15, no. 4 (Winter 2003): pp. 471-487. Levitsky, Steven and Lucan A. Way. “The Rise of Competitive Authoritarianism,” Journal of Democracy 13, No. 2 (2002): 51-65. “Liberia 2011: Monitoring the 2011 Elections in Liberia,” Ushahidi. Accessed July 2012. http://www.liberia2011.ushahidi.com. “Liberia Economic Outlook,” African Development Bank Group (2013). Accessed April 2014. http://www.afdb.org. “Liberia Election: CDC Monrovia Protest Turns Deadly,” BBC News November 7, 2011. Accessed July 2012. http://www.bbc.co.uk. “Liberia: George Weah Earns College Degree,” The Analyst June 27, 2011. Accessed July 2012. http://allafrica.com/

Electoral Violence in Sub-Saharan Africa

183

“Liberia President Ellen Johnson-Sirleaf to Stand Again,” BBC News January 26, 2010. Accessed July 2012. http://news.bbc.co.uk. “Liberia: Supreme Court Ratifies Constitutional Amendment,” US Library of Congress. Accessed July 2012. http://www.loc.gov. Lijphart, Arend. “Constitutional Choices for New Democracies,” Journal of Democracy 2, No. 1 (1991): 72-84. Lijphart, Arend. “The Problem of Low and Unequal Voter Turnout – and What We Can Do About It,” IHS Political Science Series No. 54 (1998). Lindberg, Staffan. Democracy and Elections in Africa. Baltimore: Johns Hopkins Press, 2006. Lindberg, Staffan. “The Surprising Significance of African Elections,” Journal of Democracy 17, No.1 (2006): 139–51. Lindberg, Staffan I., ed., Democratization by Elections: A New Mode of Transition. Baltimore: Johns Hopkins University Press, 2009. Lindberg, Staffan I. “Democratization by Elections? A Mixed Record,” Journal of Democracy 20, No. 3 (2009): 86-92. Long, James D., Kanynga, Karuti, Ferree, Karen E. and Clark Gibson. “Choosing Peace over Democracy,” Journal of Democracy 24, No. 3 (2013): 140155. Lynch, Gabrielle. I Say to You: Ethnic Politics and the Kalenjin in Kenya. Chicago: University of Chicago Press, 2011. Lyons, Terrence. Voting for Peace: Postconflict Elections in Liberia. Washington D.C.: Brooking Institution Press, 1998. “Macky Sall Senegal Election Win ‘Example for Africa,’” BBC News March 26, 2012 Accessed February 2014. http://www.bbc.co.uk. “Malawi Elections 2009: Statistical Based Monitoring Analysis Report,” Afrcan Elections Project (June 2009). http://www.african-elections.org. Malawi: Parliamentary and Presidential Elections 20 May 2004. EISA Electoral Observer Mission Report (2006). www.contenteisa.org. Magaloni, Beatriz and Ruth Kricheli, “Political Order and One-Party Rule,” Annual Review of Political Science 13 (2010): 123-143. “Mau Mau Uprising: Bloody History of Kenya Conflict.” BBC April 7, 2011. Accessed March 2015. http://www.bbc.com/news/uk-12997138. Mbow, Penda. “Senegal: The Return of Personalism,” Journal of Democracy 19, No. 1 (2008): 156-169. McClanahan, Paige. “Africa’s ‘Iron Lady’ Revitalizes Liberia,” The Christian Science Monitor April 12, 2011. Accessed July 2012. www.csmonitor.com. McLaughlin-Johnson, Afifya. “Liberia ICAF Update Report—March 2012” Bureau of Conflict and Stabilization Operations, U.S. Department of State. July 2012. http://www.state.gov. Meadow, Robert. “Political Violence and the Media,” Marquette Law Review 93, No. 1 (2009):231-240. Mosley, Paul. “Kenya in the 1970s: a Review Article,” African Affairs 81, No. 323 (1982): 271-277. Mosoku, Geoffrey. “Leaders Prop Mudavadi as Compromise Candidate,” The Standard, December 6, 2012. December 2012. www.standardmedia.co.ke. Mueller, Susanne D. “The Political Economy of Kenya’s Crisis,” Journal of Eastern African Studies 2, No. 2 (2008): 185-210.

184

Bibliography

  Mylonas, Harris and Nasos Roussias. “When Do Votes Count? Regime Type, Electoral Conduct, and Political Competition in Africa,” Comparative Political Studies 41, No. 11 (2008): 1466-1491. Ndegwa, Stephen N. “Kenya: Third Time Lucky?” Journal of Democracy 14 (2003): 145-158. Ndonga, Wambui. “Kenya: Militia Was Trained for Tana Killings—Human Rights Body,” AllAfrica November 28, 2012. Accessed December 2012. http://allafrica.com. “Nigeria: Post-Election Violence Killed 800.” Human Rights Watch May 17, 2011. Accessed May 2014. http://www.hrw.org. “None of the Candidate’s Garners 50%+1 for Clear Win—4 Battleground, 4 Contested Counties,” Ipsos Synovate February 22, 2013. February 2013. http://www.ipsos.co.ke. Norris, Pippa. “Why Electoral Malpractices Heighten Risks of Electoral Violence.” American Political Science Association Annual Meeting APSA 2012, New Orleans: American Political Science Association. Accessed March 2015. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=210455. Nossiter, Adam. “An Atypical Unrest Troubles Senegal’s Election Season,” The New York Times, February 23, 2012. http://www.nytimes.com. Nossiter, Adam. “A Turbulence-Free Election in Senegal,” The New York Times March 25, 2012. Accessed February 2013. http://www.nytimes.com. Nossiter, Adam. “Mali Holds Elections after Year of Turmoil,” The New York Times July 28, 2013. Accessed April 2014. http://www.nytimes.com/2013/ 07/29/world/africa/mali-holds-elections-after-year-of-turmoil.html. O’Brien, Donal Cruise. “Political Opposition in Senegal: 1960-1967.” Government and Opposition 4, No. 2 (1967): 557-566. “Observing the 2002 Kenya Elections: Final Report.” The Carter Center (2002). Accessed February 2015. https://www.cartercenter.org/documents/1355.pdf “Observing Presidential and Legislative Elections in Liberia,” National Democratic Institute (October – November 2005). Accessed April 2014. https://www.ndi.org/files/2119_lr_finalreport_110105_0.pdf. “Odinga Rejects Kenya Poll Result,” BBC News, December 31, 2007. Accessed February 2015. http://news.bbc.co.uk/2/hi/7165406.stm. Obakhedo, Neville Onebamhoi. “Curbing Electoral Violence in Nigeria: The Imperative of Political Education.” African Research Review 5, 5 (2011): 99-110. Ogana, Joseph. “Tanzania: Of CCM’s Economic Supremacy and Gangs,” The Citizen October 18, 2011. Accessed April 2015. http://allafrica.com/stories. Ombati, Cyrus. “Fresh Violence Rocks Garissa,” The Standard November 20, 2012. Accessed January 2013. http://www.standardmedia.co.ke. Ombati, Cyrus. “IEBC Top Officials Arrested over Sh1.3b Poll Kits Tender,” The Standard October 30, 2013. http://www.standardmedia.co.ke. Omolo, Ken. “Political Ethnicity in the Democratization Process in Kenya,” African Studies 61, No. 2 (2002): 209-221. Optiz, Christian, Fjelde, Hanne, and Kristine Höglund. “Including Peace: the Influence of Electoral Management Bodies on Electoral Violence,” Journal of Eastern African Studies 7, No. 4 (2013): 713-731. Oyugi, Walter O. “Ethnicity in the Electoral Process: the 1992 General Elections in Kenya,” African Journal of Political Science 2, No. 1 (1997): 4169.

Electoral Violence in Sub-Saharan Africa

185

“PARLINE Database on National Parliaments,” Inter-Parliamentary Union. Accessed April 2014. http://www.ipu.org. “Peace Initiative Kenya.” United States Agency for International Aid October 21, 2014. Accessed March 2015. http://www.usaid.gov. Phillips, Michael. “Economic Austerity Remains a Divisive Issue in Senegal.” Christian Science Monitor July 22, 1988. Accessed March 2013. http://www.csmonitor.com. “Playing with Fire: Weapons Proliferation, Political Violence, and Human Rights in Kenya,” Human Rights Watch (2002). Accessed April 2014. http://www.hrw.org/reports/2002/kenya/. Powell, G. Bingham. Elections as Instruments of Democracy. New Haven, CT: Yale University Press, 2000. Posner, Daniel N. and Daniel J. Young. “The Institutionalization of Political Power in Africa.” Journal of Democracy 18, No. 3 (July 2007): 126-140. Prempeh, H. Kwasi. “Presidents Untamed,” Journal of Democracy 19, No. 2 (2008): 109-123. “A Pre-Election Report and Advisory on Violence in Nigeria’s 2015 Elections,” National Commission on Human Rights. Accessed April 2015. http://citizenshiprightsinafrica.org. “Pre-Publication Final Report: National Elections in Liberia, Fall 2011,” The Carter Center. Accessed June 2012. http://www.cartercenter.org. “Presidential Alliances—February 2013.” Ipsos Synovate. Accessed March 2015. http://www.ipsos.co.ke. “President Sirleaf Accepts Resignation of NEC Chairman James Fromayan; Names Commissioner Elizabeth Nelson to Chair Commission,” Government of the Republic of Liberia Executive Mansion Press Release October 30, 2011. Accessed July 2012. http://emansion.gov.lr. Puddington, Arch, “The Democratic Leadership Gap.” Journal of Democracy 25, No. 2 (2014): 78-92. Raleigh, Clionadh, Linke, Andrew, Hegre, Håvard, and Joakim Karlsen. “Introducing ACLED-Armed Conflict Location and Event Data,” Journal of Peace Research 47, No. 5 (2010): 1-10. Ross, Daniel. Violent Democracy. Cambridge: Cambridge University Press, 2005. Rutten, Marcel and Sam Owuor. “Weapons of Mass Destruction: Land, Ethnicity, and the 2007 Elections in Kenya.” Journal of Contemporary African Studies 27 (2009): 305-324. Sarsfield, Rodolfo and Fabian Echegaray, “Opening the Black Box: How Satisfaction with Democracy and Its Perceived Efficacy Affect Regime Preference in Latin America,” International Journal of Public Opinion Research 18, no.2 (2005): 153-173. Sawyer, Amos. “Emerging Patterns in Liberia’s Postconflict Politics: Observations from the 2005 Elections,” African Affairs Vol. 107, 427 (2008): 177199. “Senegal: Freedom of the Press 2012,” Freedom House (2012). Accessed March 2013. http://www.freedomhouse.org/report/freedom-press/2012/senegal. “Senegal: Independence Strengthened in a Mixed Model of Electoral Management,” ACE Project. Accessed March 2013. http://aceproject.org. “Senegal Opposition Puts off Rally to Pay Tribute to Poll ‘Martyrs,’” AFP World Service March 2, 2012.

186

Bibliography

  “Senegal Violence Continues over President’s Third-Term Bid,” The Guardian UK February 19, 2012. Accessed February 2013. www.guardian.co.uk. “Senegal: Will the Momentum Last?” Open Society Foundations October 26, 2011. Accessed March 2013. www.opensocietyfoundations.org. “Senegal’s Wade Drops Contested Electoral Reform,” AFP World Service June 23, 2011. Singh, Shane, Karakoc, Ekrem, and Andre Blais, “Differentiating Winners: How Elections Affect Satisfaction with Democracy,” Electoral Studies 31 (2012): 201-211. Sisk, Timothy D. “Evaluating Election-Related Violence: Nigeria and Sudan in Comparative Perspective,” in Voting in Fear: Electoral Violence in SubSaharan Africa, ed. Dorina A. Bekoe. Washington DC: United States Institute of Peace Press, 2012. Smets, Kaat and Carolien van Ham “The Embarrassment of Riches? A MetaAnalysis of Individual-Level Research on Voter Turnout,” Electoral Studies 32, No. 1 (2013): 344-359. Snyder, Jack. From Voting to Violence: Democratization and Nationalist Conflict. New York: W.W. Norton and Company, 2000. “SPEC Barometer Survey, March 2014.” Ipsos Synovate March 4, 2014. Accessed March 2015. http://www.ipsos.co.ke. “SPEC Barometer Survey, November 2012.” Ipsos Synovate November 20, 2012. Accessed March 2015. www.ipsos.co.ke. Stafford, Andy. “Senegal: May 1968, Africa’s Revolt,” in 1968: Memories and Legacies of a Global Revolt. Phillipp Gassert and Martin Klimke, eds. Bulletin of the German Historical Institute, Supplement 6 (2009). Accessed March 2015. http://ghidc.org. Starr, Penelope. “Senegal Gripped by Pre-Election Violence,” UN Dispatch 22 February 22, 2012. Accessed February 2013. http://www.undispatch.com. Straus, Scott. “’It’s Sheer Horror Here’: Patterns of Violence during the First Four Months of Côte D’Ivoire’s Post-Electoral Crisis,” African Affairs 110 (2011): 481-489. Straus, Scott and Charlie Taylor. “Democracy and Electoral Violence in SubSaharan Africa, 1990-2008,” in Voting in Fear: Electoral Violence in SubSaharan Africa, ed. Dorina A. Bekoe. Washington DC: United States Institute of Peace Press, 2012. Sweeney, Jr., William R. “Kenya’s 2013 Elections: An Effective Assistance Model? Testimony of William R. Sweeney, Jr. to the US House Committee on Foreign Affairs.” IFES (April 16, 2013). Accessed February 2015. http://www.ifes.org. “They Killed Them Like It Was Nothing: The Need for Justice for Côte d’Ivoire’s Post-Election Crimes.” Human Rights Watch (2011). Accessed April 2014. www.hrw.org. Throup, David. “Elections and Political Legitimacy in Kenya,” Africa 63, No 3 (1993): 371-396. “The Truth and Reconciliation Commission of Liberia: Final Report,” Truth and Reconciliation Commission of Liberia (2011). Accessed July 2012. http://trcofliberia.org. Travaglianti, Manuela. Threatening Your Own: Electoral Violence within Ethnic Groups in Burundi and Beyond. (Unpublished Doctoral Dissertation). New York University, 2014.

Electoral Violence in Sub-Saharan Africa

187

Tsoka, Maxton Grant, “A Tale of Two Presidents: Assessments of ‘Chitsulo Cha Njanje’ and ‘Amayi’ in Malawi.” Afrobarometer Briefing Paper 121 (April 203). Accessed April 2014. http://www.afrobarometer.org. “UN Concerned about ‘Excessive Force’ by Senegal Authorities,” AFP North European Service February 2, 2012. “UNMIL: United Nations Mission in Liberia Facts Figures,” United Nations. Accessed July 2012. http://www.un.org. “Une Affair de Famille,” Africa Confidential April 17, 2009. Accessed March 2013. http://www.africaconfidential.com. “USAID Support for Kenya’s 2013 Elections: Rapid Assessment Review,” United States Agency for International Development (November 2013). Accessed February 2014. http://www.usaid.gov. van Cranenburgh, Oda. “Democracy Promotion in Africa: the Institutional Context.” Democratization 18, 2 (2011): 443-461. Villalon, Leonardo A. “Democratizing a (Quasi) Democracy: The Senegalese Elections of 1993,” African Affairs 93, 371 (1994): 163-193. Villalon, Leondardo A. “Senegal: Assessing Risks to Stability,” Center for Strategic and International Studies (June 2011). Accessed March 2013. http://csis.org. “Voter Turnout Database,” International IDEA. Accessed April 2013. http://www.idea.int/vt. “The Waki Report,” The Commission of Inquiry on Post-Election Violence (2008). Accessed December 2013. http://www.knchr.org. Wanyama, Frederick O., Elklit, Jorgen, Frederiksen, Bodil Folke, and Preben Kaarsholm. “Ethnicity and/or Issues? The 2013 General Elections in Western Kenya,” Journal of African Elections 13, No. 2 (2014): 169-195. “What Is the Malawi Election Situation Room: Making Elections Count,” Malawi Election Information Centre. May 2014. http://malawivote2014.org. Weidmann, Nils B. and Michael Callen. “Violence and Election Fraud: Evidence from Afghanistan,” British Journal of Political Science 43, No. 1 (2013): 53-75. “World Development Indicators.” World Bank. July 2012. http://databank.world bank.org.

Index

Angola, 3, 18, 24, 50, 51, 123 Autonomous National Electoral Commission (CENA), 91, 97 Autocratic elections, 43, 52, 126-127, 142, 150, 163, 170-171 Benin, 2, 32-33, 36, 141, 151 Boko Haram, 14 Botswana, 2, 36-37, 69, 125, 130, 141, 149; democratic attitudes in: 151, 153, 155 Burkina Faso, 50, 93, 104, 141, 171 Burundi, 30, 36, 50, 123, 125-126, 170 Cameroon, 36, 50, 93, 153-154 CAR. See Central African Republic Carter Center, 2, 79, 107-109, 115, CDC. See Congress for Democratic Change CENA. See Autonomous National Electoral Commission Central African Republic (CAR), 172 Colonialism: in Kenya, 56-58; in Senegal, 87-88 Compaoré, Blaise, 171 Congress for Democratic Change (CDC), 108-111, 113-115, 116117, 119 Constitutional Reform, 4, 7, 59, 7475, 77, 91-94, 171 Côte d’Ivoire, 3, 35-36, 50, 104, 123, 143 Crisis mapping platforms: Uchaguzi, 78; Ushahidi, 116 Democracy, 2, 7, 11, 18-20, 41-43, 48, 50-51, 85-86, 104, 123-124, 143-144, 147-152, 164, 172-173 Democratic Republic of Congo (DRC), 2, 24, 36, 50, 123, 128, 130, 171 Dia, Mamadou, 88-89

Diouf, Abdou , 7, 90-91, 94, 96, 97 Doe, Samuel K., 9, 103-105, 109, 111 DRC. See Democratic Republic of Congo ECK. See Electoral Commission of Kenya ECOMOG. See Economic Community of West Africa Monitoring Group Economic Community of West Africa Monitoring Group (ECOMOG), 106 Eldoret, Kenya, 1, 5, 63 Election boycott: in Liberia, 9, 104, 111, 113-115, 117; in Senegal: 89, 90-92 Election management, 7, 25-26, 34, 41, 96-99, 115, 161, 168-169, 171 Election monitoring, 34-35, 41, 43, 51, 75, 78-80, 96-99, 115, 117118 Electoral Commission of Kenya (ECK), 6, 25, 62-63, 75 Electoral fraud, 6, 9, 16, 23, 26-27, 150, 152, 169-170 Electoral rules, Majoritarian rules, 26, 30-34, 38, 40, 44, 47-48, 94, 108,113, 168; Mixed systems, 31, 33, 40, 44, 47-48, 168; Plurality, 30-34, 3740, 44-48, 73, 75, 106, 108, 113, 168-169; Proportional representation, 26, 30-34, 39-40, 44, 48, 106; Ethiopia, 1, 24, 25, 95, 125, 170 Ethnicity: in Kenya, 4-5, 32, 56-61, 65, 69-73; in Liberia, 103, 105; in Senegal, 85 Ghana, 69, 125, 149, 153-154

189  

190

Index

ICC. See International Criminal Court IEBC. See Independent Election and Boundaries Commission IFES. See International Foundation for Electoral Systems International Criminal Court (ICC), 76, 82, 110 Independent Election and Boundaries Commission (IEBC), 75, 77, 7981 International Foundation for Electoral Systems (IFES), 2, 79, 112, 128 Judiciary, 43, 74-75, 80, 118, 170 KANU. See Kenya African National Union KAU. See Kenya African Union. Kenya African National Union (KANU), 2, 5, 28, 55, 57-62, 7172, 74, 78 Kenya African Union (KAU), 56-57 Kenya People’s Union (KPU), 4, 58 Kenyatta, Jomo, 1, 4-5, 55-56, 58-59 Kenyatta, Uhuru, 2, 62, 71, 74, 76-78 Kibaki, Mwai, 4, 16, 65, 66, 73, 7475; in 1997 election, 61; in 2002 election, 62, 75; in 2007 election, 1-2, 4-7, 16, 62-65, 67-71, 79 Kisumu, Kenya, 1, 55, 63, 77 Kivuitu, Samuel, 6, 63 KPU. See Kenya People’s Union Lesotho, 33, 51, 69, 141, 153-154 Liberian Civil War, 9, 104-106, 136 Madagascar, 50, 141, 153-155 Malawi, 25, 69, 93, 153-154 Mali, 50, 69, 125, 141, 153-154, 172 Mau Mau Rebellion, 56-57 Mboya, Tom, 55, 58 Mkapa, Benjamin, 74 Moi, Daniel arap, 1, 5, 14, 28, 31-32, 56, 58-62, 71, 74 Molo, Kenya, 1, 5-7 Mombasa, Kenya, 1, 63, 66 Mombasa Republican Council (MRC), 14 MRC. See Mombasa Republican Council Mugabe, Robert, 153, 170 Mungiki, 28, 136 Museveni, Yoweri, 170

NARC. See National Alliance Rainbow Coalition National Alliance Ranibow Coalition (NARC), 62, 71-72, 74, 78 National Elections Commission (NEC), 107-108, 112-115 NEC. See National Elections Commission Nigeria, 2, 9, 12, 14, 23-24, 28, 29, 36, 69, 86, 95, 141, 149, 153-154, 172; in the Liberian Civil War, 106-107; qualified majority electoral rules, 32, 51; term limits, 93; the 2007 election, 125126; the 2011 election, 26-27, 117, 125, 149 Odinga, Jaramogi Oginga, 4-5, 58-59 Odinga, Raila, 1, 4-5, 14, 58, 61-62, 71, 74-77 ODM. See Orange Democratic Movement Orange Democratic Movement (ODM), 4-6, 14, 63, 71-72, 76, Partisanship, 18, 70-71, 130, 137140, 155, 164 Party of National Unity (PNU), 63, 71-72 Party switching, 48 PDS. See Senegalese Democratic Party PEV. See Post-Election Violence in Kenya PNU. See Party of National Unity Political parties, 32-33, 40, 48, 71-73, 106, 108, 127, Post-Election Violence in Kenya (PEV), 1-2, 4-6, 27, 62-70, 73, 76, 141, 160, 163 PS. See Socialist Party Republic of Congo, 143 Rift Valley Province, Kenya, 4, 5-6, 14, 57-63, 65, 69, 73, 75 Rule of law, 2, 11, 37, 43-44 Ruto, William, 5, 75-77 Rwanda, 30, 32, 149, 170, 171 Sall, Macky, 91, 93-95 Seck, Idrissa, 91-93 Senegalese Democratic Party (PDS), 89-91 Senghor, Léopold Sédar, 7, 88-91

Electoral Violence in Sub-Saharan Africa 191

Sirleaf, Ellen Johnson, 9, 107-112, 114-119 Socialisty Party (PS), 7, 89-90, 97 South Africa, 29-30, 32, 37, 69, 153154, 170 Sudan, 1, 36, 130, 172 Tanzania, 28, 36, 69, 74, 153-154, 170 Taylor, Charles, 9, 104-110, Term limits, 92-93, 170; in Burkina Faso; in in Kenya, 1, 61, 92 ; in Liberia, 119; Senegal, 7, 92-94, 96 Togo, 24, 36, 50, 128, 154 Tolbert, William, 9, 105, 109 True Whig Party (TWP), 9, 103, 105 TWP. See True Whig Party Uganda, 1, 36, 69, 93, 123, 153-154, 170 UN. See United Nations United Nations (UN), 9, 104, 107110 United Nations Mission in Liberia (UNMIL), 9, 118 Unity Party (UP), 108, 111, 113-114, 116-117, 119 UNMIL. See United Nations Mission in Liberia UP. See Unity Party Wade, Abdoulaye, 7-8, 86, 89-99, 137, 169 Wade, Karim, 91-92 Waki Commission and Report, 6, 14, 63 Weah, George, 9, 108-111, 119 Y’en A Marre, 93 ZANU-PF. See Zimbabwe African National Union-Patriotic Front Zambia, 31, 69, 93, 141, 149, 153154, 170 Zimbabwe, 2, 12, 23-24, 28, 50, 69, 85-86, 94-95, 123, 125, 141, 153155, 170 Zimbabwe African National UnionPatriotic Front (ZANU-PF), 28, 153

About the Book

After decades of experimentation with various forms of dictatorship and autocracy, most sub-Saharan African countries adopted multiparty elections in the 1990s—a development widely celebrated as a sign that the region was moving toward democracy. This embrace of elections, however, has often been accompanied by unanticipated violence, raising important questions: Are violent elections a normal part of the process in new democracies? Does the quality and conduct of elections matter for democratic consolidation? Most fundamentally, what does the persistence of electoral violence mean for the future of democracy in Africa? Addressing these questions with a combination of rigorous qualitative and quantitative approaches, Stephanie Burchard explores both the causes and consequences of electoral violence in sub-Saharan Africa. Stephanie M. Burchard is on the research staff of the Africa Program at the Institute for Defense Analyses.

193