THE CAUSES AND CONSEQUENCES OF ELECTORAL ADMINISTRATIVE REFORM IN AFRICA By Nicholas N. Kerr A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Political Science — Doctor of Philosophy 2013 ABSTRACT THE CAUSES AND CONSEQUENCES OF ELECTORAL ADMINISTRATIVE REFORM IN AFRICA By Nicholas N. Kerr In emerging democratic settings, scholars and policy experts regard electoral management bodies (EMBs) as one of the most important institutions in guaranteeing that political actors adhere to the rules of the electoral contest and that the substantive outcome of elections is not predetermined. Unfortunately, systematic and thorough assessment of EMBs has largely escaped the spotlight of democracy scholars. My dissertation probes two critical questions concerning the important, yet understudied role of electoral management bodies in emerging democracies. First, what factors influence the design of electoral administration in Africa? Second, what are the consequences of electoral administrative design and performance on the quality of elections in Africa? To answer the first question, I propose a theoretical framework, grounded in historical institutionalism, which attributes election administrative reform to the strategic interactions of domestic and international actors along with the path dependency of institutional legacies during different stages of democracy. Through a comparative analysis of election administrative reforms in a cross-section of African countries, I emphasize the complexity of the reform process and highlight various models of institutional reform. I argue, however, that incumbents are more likely to enact reforms that enhance the autonomy and capacity of electoral administration when they face credible and enduring pressure from a coalition of opposition parties, civil society groups and international actors. In subsequent chapters, I turn my attention to the consequences of electoral administrative design and performance. Primarily, I consider how EMB design and performance may shape the incentives and behaviors of political elite and regular citizens. Using crossnational data on election administrative design and performance in Africa from 1990 to 2008 and data on elites’ and citizens’ perspectives of election quality, I find EMB performance has the greatest impact on electoral attitudes and behavior. That is, EMBs that display higher levels of actual autonomy and capacity are more likely to enhance electoral integrity by reducing opportunities for electoral fraud, lowering the incidence of procedural irregularities and enhancing citizens’ quality of participation in electoral processes. In the penultimate chapter of the dissertation, I rely on survey data from the 2007 and 2011 Nigerian elections to consider whether citizens’ personal evaluations of electoral administrative autonomy and capacity are positively associated with their perceptions of election quality. The findings confirm my expectations, even after accounting for Nigerians’ experience with electoral irregularities and their political preferences. I use a variety of methods in my dissertation. These include large-n statistical analysis of survey data and other data, as well as small-n comparative analysis of election management in Ghana, Nigeria, South Africa and Zimbabwe using process tracing and content analysis. This multi-method approach also includes collecting original cross-national data on election administrative design in Africa, conducting elite interviews in Ghana, Nigeria, South Africa and Zimbabwe, and observing elections in Ghana (2008) and Nigeria (2011). Copyright by NICHOLAS N. KERR 2013 ACKNOWLEDGEMENTS This dissertation represents over four years of research while I was a student in the Political Science Department of Michigan State University that could not have been possible without the support and encouragement of family, friends and colleagues. While it would be difficult to mention everyone, I would like to briefly acknowledge those who were pivotal in the process. The chair of my doctoral dissertation, Michael Bratton, was a tireless supporter of the project since its inception. Not only was he convinced that my crazy idea of conducting fieldwork in four countries was doable but at every step of the process, he provided instructive comments on my work, invaluable advice and encouragement that made all the difference especially at times when I found myself on the verge of throwing in the towel. I am equally indebted to Eric Chang, Rita Edozie and Thomas Hammond who served as members of the dissertation committee provided constructive feedback and vital mentorship. I appreciatively acknowledge the support of other members of the MSU community. These include the faculty and staff within the Political Science Department, such as Carolyn Logan (and the Afrobarometer), Jeffery Conroy-Krutz, Steve Kautz, Karen Battin, Rhonda Burns and Sarah Krause. I was fortunate to have been surrounded by fellow graduate studentsTwyla Blackmond-Larnell, Danielle Carter, Seyoun Choi, Brian Kennedy, Jerry Lavery and Helen Leewhom I could rely on to provide feedback on drafts, to help me prepare for conferences, to resolve methods inquiries or to get a burger at Peanut Barrel. The project’s success hinged upon the completion of fieldwork in Ghana, Nigeria, South Africa and Zimbabwe. This was only feasible because of the unrequited generosity of numerous v individuals. In Ghana, I am thankful for the gracious support of Professor Gyimah-Boadi (Center for Democracy and Development-Ghana), the resourcefulness of James Arthur-Yeboah, and the hospitality of Emmanuel Treku. In Nigeria, institutional affiliations with Center of Democracy and Development-West Africa (through Professor Jibrin Ibrahim) and Redeemers University opened many doors; special thanks also to Bodunrin Adebo, Covenant Child, Simon Fanto, Lloyd Jackson, and Shola Omotola. The South African leg of the project was possible through the assistance of Robert Mattes and the staff at the Democracy and Africa Research Unit at University of Cape Town. In Zimbabwe, I was blessed to have been affiliated with Mass Public Opinion Institute (through Professor Eldred Masunungure). I also received invaluable assistance from Tawanda Chimhini and Jabusile Shumba. In the USA, I would be remiss not to thank those who provided assistance in planning the various fieldwork stints, such as Gemima Barlow, Keith Jennings, Daren Kew, Richard Klein, Oge Okoye-Onubogu, and Peter Smith. Thanks very much for the vital contribution that you all played in making this process a reality. The dissertation would not have gotten off the ground without the financial support of MSU through its pre-dissertation fellowship and graduate enhancement awards, the International Foundation of Electoral Studies through the Manatt Research Fellowship, the Democracy and Africa Research Unit at University of Cape Town through its Research Writing Fellowship, and the Afrobarometer through numerous awards and grants. To my family, especially my Mother, Father and Uncle Clyde, thank you for helping me realize that “ with God all things are possible” and for your countless displays of love and support. Mommy, your own Ph.D. journey and unwavering faith provided the perfect blueprint during the dissertation process. Dad, your constant phone calls and frequent trips to East Lansing were the perfect distraction from the stress of writing. Uncle Clyde, you continue to be my vi spiritual compass and the consummate listening ear regardless of the time or subject. I also thank my extended family and friends whose prayers and frequent words of encouragement have not gone unnoticed. To my wife Shani, I dedicate this dissertation to you. I found in you the most supportive wife anyone could have asked for. Your selfless commitment to helping me complete this journey can only be matched by your patience, thoughtfulness, constant prayers, and irresistible jerk pork and rice and peas. During the project you filled multiple roles: sounding board, data coder, editor, financial manager, and telephone counselor/nurse during the bouts of malaria and typhoid in the field. You are truly a “phenomenal” woman of God. Finally, and most importantly I thank God for helping me to completely understand that “all things work together for good to them that love God and to them that are called according to his purpose” (Romans 8:28). vii TABLE OF CONTENTS LIST OF TABLES ………………………………………………………………………….. xiii LIST OF FIGURES …………………………………………………………………………. xvi CHAPTER 1 INTRODUCTION …………………………………………………………………………... 1.1 Introduction ………………………………………………………………………... 1.2 Main Arguments …………………………………………………………………… 1.2.1 Causes of EMB Design ……………………………………………………… 1.2.2 Consequence of EMB Design and Performance ……………………………. 1.2.3 Linking Causes and Consequences …………………………………………. 1.3 What is Electoral Administration? …………………………………………………. 1.3.1 Electoral Management Bodies ………………………………………………. 1.4 Why Examine Electoral Administration? …………………………………………. 1.5 Why Africa? ………………………………………………………………………... 1.6 Theoretical Approach ……………………………………………………………… 1.6.1 Formal and Informal Political Institutions …………………………………… 1.6.2 Institutional Design and Performance ……………………………………….. 1.6.3 Dimensions of Institutionalization: Autonomy and Capacity ……………….. 1.6.4 Historical Institutionalism …………………………………………………… 1.7 Research Methods ………………………………………………………………….. 1.8 Research Contributions …………………………………………………………….. 1.9 Roadmap of Subsequent Chapters ………………………………………………… APPENDIX ……………………………………………………………………………. 1 1 3 3 4 6 6 7 8 11 11 12 13 14 16 19 20 21 25 CHAPTER 2 DATA AND METHODS …………………………………………………………………… 2.1 Introduction ………………………………………………………………………… 2.2 Existing Data Sources on EMB Design and Performance …………………………. 2.2.1 Data on EMB Design ………………………………………………………… 2.2.2 Data on EMB Performance …………………………………………………... 2.3 Constructing a Dataset on EMB Design in Africa …………………………………. 2.3.1 EMB Autonomy ……………………………………………………………… 2.3.2 EMB Capacity ……………………………………………………………….. 2.4 Additional Large-n Data Sources ………………………………………………….. 2.5 Research Methods Small-n Analysis ………………………………………………. 2.5.1 Case Selection ………………………………………………………………... 2.5.2 Semi-Structured Interviews ………………………………………………….. 2.5.3 Participant Observation ……………………………………………………… APPENDIX …………………………………………………………………………..... 32 32 32 32 34 35 36 37 38 39 40 40 42 43 viii CHAPTER 3 THE CAUSES OF ELECTORAL ADMINISTRATIVE REFORM IN AFRICA …………. 3.1 Introduction ………………………………………………………………………… 3.2 Literature Review and Theory ……………………………………………………... 3.2.1 Historical Institutionalism: Explaining Institutional Change ………………..... 3.2.2 Explaining EMB Reform: A Historical Institutionalism Approach …………... 3.2.3 Explaining EMB Reform: Theoretical Framework ………………………….... 3.2.4 Impact of EMB Reform on an Incumbent’s Electoral Strategy ……………… 3.2.5 Strength of Opposition ……………………………………………………....... 3.2.6 Incumbent Organizational and Coercive Power ……………………………… 3.2.7 Role of International Community …………………………………………….. 3.2.8 Critical Junctures and Path Dependency ……………………………………… 3.3 Data and Methods ………………………………………………………………….. 3.3.1 Qualitative Methodology ……………………………………………………... 3.3.2 Measuring EMB Reform ……………………………………………………… 3.3.3 Domestic and International Balance of Power ……………………………….. 3.3.4 Critical Junctures ……………………………………………………………… 3.4 Empirical Findings ………………………………………………………………… 3.4.1 Ghana (1992-2012): A Model for Africa …………………………………….. 3.4.2 South Africa (1994-2009): Forerunner in Electoral Administrative Innovation 3.4.3 Nigeria (1999-2011): Getting Election Administration Right in 2012, After Three Unsuccessful Attempts. ……………………………………………….. 3.4.4 Zimbabwe (2000-2008): Significant EMB Reforms No Match for the Longstanding Incumbent ……………………………………………………. 3.5 Discussion ………………………………………………………………………….. 3.6 Conclusion …………………………………………………………………………. APPENDIX …………………………………………………………………………… CHAPTER 4 THE CONSEQUENCES OF EMB DESIGN ON ELITES AND CITIZENS IN AFRICA 4.1 Introduction………………………………………………………………………… 4.2 Literature Review ………………………………………………………………….. 4.2.1 Effect of EMB Autonomy on Elite Attitudes ………………………………… 4.2.2 Effect of EMB Autonomy on Citizen Attitudes ……………………………… 4.2.3 Effect of EMB Capacity on Citizens and Elites ……………………………… 4.3 Consequences of EMB Design on Elite Views of Election Quality ……………….. 4.3.1 Dependent Variable: Overall Election Quality ……………………………….. 4.3.2 Independent Variable: EMB Design ………………………………………….. 4.3.3 Controls ……………………………………………………………………….. 4.3.6 Results ………………………………………………………………………… 4.4 Consequences of EMB Design on Citizen Views of Election Quality …………….. 4.4.1 Dependent Variable: Citizens’ Election Quality Evaluations………………… 4.4.2 Independent Variables: EMB Design …………………………………………. 4.4.3 Election-level Controls ………………………………………………………... 4.4.4 Individual-level Controls ……………………………………………………... 4.4.5 Results ………………………………………………………………………… ix 49 49 55 55 58 61 61 63 65 65 67 68 69 69 70 71 71 71 75 79 84 91 93 95 109 109 113 114 117 118 120 121 122 126 127 129 130 132 133 133 134 4.5 Discussion ………………………………………………………………………….. 135 4.6 Conclusion …………………………………………………………………………. 137 APPENDIX …………………………………………………………………………… 138 CHAPTER 5 THE CONSEQUENCES OF EMB PERFORMANCE ON ELITES’ ELECTORAL ATTITUDES AND BEHAVIOR …………………………………………………………... 5.1 Introduction ………………………………………………………………………… 5.2 Consequences of EMBs on Elites: Theoretical Approach. ………………………… 5.2.1 Historical Institutionalism …………………………………………………….. 5.2.2 Characteristics of EMBs: Autonomy and Capacity …………………………... 5.2.3 EMB Autonomy ………………………………………………………………. 5.2.4 EMB Capacity ………………………………………………………………… 5.2.5 Sources of EMB Autonomy and Capacity …………………………………… 5.2.6 Observable Implications and Hypotheses ……………………………………. 5.3 Consequences of EMBs for Elites in Ghana, Nigeria, South Africa and Zimbabwe 5.3.1 Role of EMBs in Elections …………………………………………………… 5.3.2 EMB Autonomy and Capacity ……………………………………………….. 5.3.3 Gap between EMB Design and Performance ………………………………… 5.4 Consequences of EMBs for Elite Attitudes and Behavior: A Statistical Analysis … 5.4.1 Consequences for EMB Performance (De Facto Autonomy and Capacity) 5.4.2 Main Independent Variable: EMB Performance (De Facto autonomy and capacity) ……………………………………………………………………… 5.4.3 Determinants of Overall Election Equality …………………………………… 5.4.4 Determinants of Election Boycotts …………………………………………… 5.4.5 Determinants of Losers’ Acceptance of Results ……………………………… 5.4.6 Determinants of Electoral Turnover ………………………………………….. 5.5 Discussion …………………………………………………………………………. 5.6 Conclusion …………………………………………………………………………. APPENDIX …………………………………………………………………………… CHAPTER 6 THE CONSEQUENCES OF EMB PERFORMANCE ON CITIZENS’ ELECTION QUALITY EVALUATIONS ……………………………………………………………….. 6.1 Introduction: Give us a Competent and Unbiased Electoral Umpire!……………… 6.2 Literature Review ………………………………………………………………….. 6.2.1 Conceptualizing Election Quality Perceptions ……………………………….. 6.2.2 Sources of Citizens Election Quality Perceptions ……………………………. 6.2.3 Institutional Performance …………………………………………………….. 6.2.4 Characteristics of the Electoral Environment ………………………………… 6.2.5 Citizens’ Individual Attributes ……………………………………………….. 6.3 How Do Citizens Evaluate Election Quality in Africa? …………………………… 6.3.1 What Types of Elections Do Citizens Want? ………………………………… 6.3.2 How do EMBs Help Promote Free and Fair Elections? ……………………… 6.3.3 EMB Characteristics: Autonomy and Capacity ………………………………. 6.3.4 EMB Autonomy ………………………………………………………………. x 155 155 158 158 159 160 165 166 169 169 170 172 173 175 175 175 177 179 180 181 182 184 185 192 192 196 196 197 197 198 200 203 203 203 205 206 6.3.5 EMB Capacity ………………………………………………………………… 6.3.6 How do Citizens Evaluate EMB Autonomy and Capacity? ………………… 6.3.7 Observable Implications and Hypotheses …………………………………… 6.4 Descriptive Statistics……………………………………………………………….. 6.4.1 Africans’ Evaluations of Election Quality in Africa ………………………….. 6.4.2 Africans’ Evaluations of Election Quality by Country ……………………….. 6.4.3 Africans’ Evaluations of Election Quality Compared to Other Regions of the World …………………………………………………………………………. 6.4.4 Africans’ Trust in EMBs ……………………………………………………. 6.4.5 Africans’ Awareness of EMBs ……………………………………………… 6.5 Consequences of EMBs on Citizens’ Election Quality Evaluations: Statistical Analysis…………………………………………………………………………….. 6.5.1 Dependent Variable: Citizens’ Election Quality Evaluations ………………… 6.5.2 Independent Variables: EMB Performance …………………………………... 6.5.3 Election-level and Individual-Level Controls ………………………………… 6.5.4 Multiple Regression Results ………………………………………………….. 6.5.5 Robustness Checks and Alternative Model Specifications …………………… 6.6 Discussion ………………………………………………………………………….. 6.7 Conclusion …………………………………………………………………………. APPENDIX……………………………………………………………………………. CHAPTER 7 THE CONSEQUENCES OF EMB PERFORMANCE ON CITIZENS’ ELECTION QUALITY EVALUATIONS: A CASE STUDY OF NIGERIA (1999-2011) ……………... 7.1 Introduction………………………………………………………………………… 7.2 Elections and Election Management in Nigeria (1999-2011) ……………………... 7.2.1 Elections in Nigeria 1999-2011 ………………………………………………. 7.2.2 1998-1999 Elections: Rigged, but Better than a Return to Military Rule ……. 7.2.3 2003 Elections: Massive Fraud and Opposition Boycotts ……………………. 7.2.4 2007 Elections: “Worst elections that ever took place in the country” (TMG) 7.2.5 2011 Elections: “Substantial improvement” but “far from perfect” ………… 7.3 Nigerians’ Evaluations of the Independent National Electoral Commission (INEC) 7.3.1 Nigerians’ Trust in INEC …………………………………………………….. 7.3.2 Nigerians’ Evaluations of INEC Performance ………………………………. 7.3.3 Nigerians’ Evaluation of INEC Autonomy and Capacity ……………………. 7.3.4 Evaluations of INEC Autonomy (2007 Elections) ……………………………. 7.3.5 Evaluations of INEC Capacity (2007 Elections) ……………………………... 7.3.6 Variations in Citizens’ Evaluations of INEC Capacity and Autonomy……….. 7.4 Consequences of INEC Performance on Nigerians ………………………………... 7.4.1 Are Nigerians’ Evaluations of EMB Performance Associated with Perceptions of Election Quality? ………………………………......................... 7.4.1.1 Data and Measurement ………………………………............................ 7.4.1.2 Results …………………………………………………………………. 7.4.2 Consequences of INEC Autonomy and Capacity on Nigerians’ Election Quality Evaluations ………………………………………………………………… 7.4.2.1 Results ………………………………………………………………… xi 207 209 211 211 211 213 214 214 216 219 219 219 220 221 223 224 225 227 240 240 243 243 244 246 248 251 255 255 256 257 257 259 261 262 262 262 268 270 272 7.4.3 Comparing the 2007 and 2011 Elections ……………………………………... 7.5 Conclusion …………………………………………………………………………. APPENDIX …………………………………………………………………………… 275 280 281 CHAPTER 8 CONCLUSION ……………………………………………………………………………... 8.1 Summary of Findings …………………………………………………………... 8.2 Theoretical Implications ………………………………………………………… 8.3 Policy Implications ……………………………………………………………… 8.4 Research Limitations and Opportunities for the Future ………………………… 306 307 310 312 313 BIBLIOGRAPHY ………………………………………………………………………….. 316 xii LIST OF TABLES Table 1.1: Levels of Electoral Governance.......................................................................... 26 Table 1.2: Distribution of EMB Autonomy and Capacity ................................................... 27 Table 1.3: Distribution of EMB Models in Africa ………………………………………...28 Table 2.1: Distribution of EMB Autonomy and Capacity by Election Periods................... 44 Table 2.2: Research Sites and Number of Elite Interviews ................................................. 45 Table 3.1: Reform Choices to Enhance EMB Autonomy and Capacity.............................. 96 Table 3.2: Distribution of EMB Autonomy and Capacity by Election Periods................... 98 Table 3.3: Dimensions of Autonomy and Capacity ............................................................. 99 Table 3.4: EMB Autonomy and Capacity Reforms and Main Causal Factors by Election Period................................................................................................................... 99 Table 3.5: EMB Autonomy and Capacity Reforms in Ghana (1992-2012) ........................ 100 Table 3.6: EMB Autonomy and Capacity Reforms in South Africa (1994-2009) .............. 103 Table 3.7: EMB Autonomy and Capacity Reforms in Nigeria (1999-2012) ....................... 105 Table 3.8: EMB Autonomy and Capacity Reforms in Zimbabwe (2000-2008).................. 107 Table 4.1: Factor Analysis of EMB De Jure Autonomy Indicators .................................. .139 Table 4.2: Factor Analysis of EMB De Jure Capacity Indicators ..................................... .140 Table 4.3: Determinants of Election Quality (19-country sample).................................... .141 Table 4.4: Election Periods, Countries and Afrobarometer Rounds .................................. .142 Table 4.5: Citizens’ Evaluations of Election Quality: 28 Election Periods ....................... .143 Table 4.6: Determinants of Citizens’ Election Quality Evaluations .................................. .144 Table 5.1: Country and Election Years- 45 Country Analysis .......................................... .190 xiii Table 5.2: Correlates of Election Quality (45-country sample) ......................................... .191 Table 6.1: Africans’ Evaluation of Election Quality, 1999-2008 ...................................... .228 Table 6.2: Africans’ Evaluation of Election Quality Across 20 Countries, 1999-2008 .... .229 Table 6.3: Citizens’ Evaluations of Election Quality Across 4 World Regions ................ .230 Table 6.4: Election Periods, Countries and Afrobarometer Rounds .................................. .231 Table 6.5: Citizens’ Evaluations of Election Quality: 28 Election Periods ....................... .232 Table 6.6: Determinants of Citizens’ Election Quality Evaluations .................................. .233 Table 6.7: Predicted Probability of Free and Fair Election Quality Evaluations ……..234 Table 6.8: Determinants of Citizens’ Election Quality Evaluations (Alternative Models) 235 Table 7.1: Nigeria Election Dates (1964-2011) ........................................................ ……282 Table 7.2: Nigeria Presidential Election Results (1999-2011) ......................................... 283 Table 7.3: Nigeria National Assembly Election Results (1999-2011) ............................. 284 Table 7.4: Nigeria Registration and Turnout (1999-2011) ............................................... 285 Table 7.5: Nigeria Election Violence (1999-2011) ........................................................... 286 Table 7.6: Nigeria Election Petitions and Boycotts (1999-2011) ..................................... 287 Table 7.7: Nigeria EMB Reform (1999-2011) ................................................................. 288 Table 7.8: INEC Chairpersons (1999-2011) ..................................................................... 289 Table 7.9: Nigerians’ Evaluation of INEC Performance (2007-2011) ............................. 290 Table 7.10: INEC Autonomy (2007) ................................................................................ 291 Table 7.11: INEC Capacity (2007-2011) .......................................................................... 292 Table 7.12: Nigerians Election Quality Perceptions (2007) ............................................. 293 Table 7.13: Sources of Citizens’ Perceptions of Election Quality (2008: AB) ………….294 Table 7.14: Predicted Probability of Free and Fair Election Quality Perceptions (2008: AB)……………………………………………………………………295 xiv Table 7.15: Sources of Citizens’ Perceptions of Election Quality (2007: IFES) ……......296 Table 7.16: Predicted Probability of Free and Fair Election Quality Perceptions (2007: IFES)………………………………………………………………... 297 Table 7.16: Nigerians Election Quality Perceptions (2007 vs. 2011) by Geo-Political Zones…………………………………………………………………………298 xv LIST OF FIGURES Figure 1.1: Theoretical Model: Causes of EMB Design and Performance ......................... 29 Figure 1.2: Theoretical Model: Consequences of EMB Design and Performance .............. 30 Figure 1.3: Cycle of EMB Reform and Election Quality .................................................... 31 Figure 4.1: Theoretical Model: Causes of EMB Design and Performance ....................... 145 Figure 4.2: Mean EMB De Jure Autonomy and Capacity by Overall Election Quality 146 Figure 4.3: Citizens’ Evaluations of Election Quality by Election Period ........................ 147 Figure 4.4: Variations in EMB Design by Election Quality .............................................. 148 Figure 5.1: Mean EMB Performance by Overall Election Quality ................................... 186 Figure 5.2: Mean EMB Performance by Opposition Participation.................................... 187 Figure 5.3: Mean EMB Performance by Losers’ Acceptance of Results .......................... 188 Figure 5.4: Mean EMB Performance by Electoral Turnover ............................................ 189 Figure 6.1: Theoretical Model: Consequences of EMB Design and Performance ............ 236 Figure 6.2: Citizens’ Evaluations of Election Quality by Election Period ........................ 237 Figure 6.3: Variations in EMB Performance ..................................................................... 238 Figure 6.4: Effect of EMB Performance on Citizens’ Election Quality Evaluations ........ 239 Figure 7.1: Percentage of Citizens who Trust INEC “A lot or Somewhat” (1999-2011) 299 Figure 7.2: INEC Autonomy/Capacity by Geo-Political Zones (2007) ............................ 300 Figure 7.3: INEC Autonomy/Capacity by Partisanship (2007) ......................................... 301 Figure 7.4: INEC Autonomy/Capacity by Education (2007) ............................................ 302 Figure 7.5: INEC Autonomy/Capacity by violence Experience (2007) ............................ 303 xvi Figure 7.6: Nigerians Election Quality Perceptions by States (2007) ............................... 304 Figure 7.7: Impact of INEC Autonomy and Capacity on Predicted Probability of Free and Fair Election Quality Perceptions ……………………………………………305 xvii CHAPTER 1 INTRODUCTION 1.1 Introduction On behalf of President Obama and the people of the United States, I applaud the people of Nigeria for their enthusiastic and orderly participation in the April 16th presidential election. This historic event marks a dramatic shift from decades of failed elections and a substantial improvement over the 2007 presidential election [. . .] We commend the Independent National Electoral Commission and Chairman Professor Attahiru Jega along with many others across government and civil society for their strong collaboration and dedication to democracy (Clinton 2011). This press statement made by Secretary of State Hillary Rodham Clinton on April 19, 2011 to Nigerians echoed the sentiments of most domestic and international stakeholders, that the 2011 presidential elections were one of the most credible polls the country experienced since the reintroduction of multiparty rule in 1999 (Akhaine 2012; EU EOM 2011). It was certainly more free and fair than the ill-fated 2007 Federal and State elections, which were characterized by a leading domestic observation group as the “worst elections that ever took place in the country” (TMG 2007). What happened between 2007 and 2011 that allowed Nigerians to experience one of the most democratic elections since the reintroduction of multiparty rule in 1999? Among the various explanations, Nigerians, and members of the international community, including Secretary of State Clinton, highlighted the improved performance of the Independent National Election Commission (INEC): the institution with the constitutional responsibility for administering Nigeria’s Federal and State elections (Constitution of Nigeria 1999). During the 1 run-up to the 2011 elections, INEC underwent important reforms that helped enhance its autonomy from executive control and its capacity to effectively organize elections. These reforms included enhanced institutional and financial autonomy of INEC secured through a constitutional amendment, the appointment of a new INEC chairman with a reputation for political neutrality, a significantly expanded INEC budget (US$ 580 million), and a new voter registration system (Economist 2011; Lewis 2011; Akhaine 2012). Nigeria’s historic 2011 elections reflects a larger trend in Africa and other developing democracies where electoral commissions, formally known as electoral management bodies (EMBs), play a crucial role in shaping the integrity of elections. EMBs may be regarded as instruments of democracy: they help to ensure the procedural legitimacy and substantive uncertainty of the most crucial period in the democratic process. When EMBs perform effectively, they help create a level electoral playing field by making sure that political actors play by the rules and by reducing the extent of electoral administrative problems. Because of the important role of EMBs during elections, political actors are often engaged in battles over the institutional rules that determine the autonomy and capacity of the EMBs (Gazibo 2006). In some countries, such as Ghana, reformers have been able to secure victories through important changes in the design of EMBs, but in others incumbents have been able to maintain the status quo. However, even when reforms are passed, the ability of EMBs to improve the quality of elections by reducing executive manipulation of elections and reducing problems with management of elections is not guaranteed. The performance of EMBs depends on the effective enforcement of formal rules, influence of informal rules as well as the prevailing political and institutional contexts. 2 My dissertation examines two critical questions concerning the important yet understudied role of electoral management bodies in emerging democracies: 1. What factors influence the design and reform of electoral administrative bodies? 2. What are the consequences of electoral administrative design and performance on election quality? These two questions highlight the two main parts of the dissertation. In part 1, I examine the processes, actors, and institutional contexts that shape the institutional design of EMBs and facilitate institutional reform. In part 2, I probe the various ways through which EMBs influence the attitudes and behavior of political elites and everyday citizens. 1.2 Main Arguments 1.2.1 Causes of EMB Design To answer the first question, I propose a theoretical framework, grounded in historical institutionalism, which associates the design of election administration with 1) the strategic interaction of domestic elites (incumbent and opposition) and international actors; 2) the path dependency of existing institutional rules; and, 3) and, the characteristics of the political and institutional context. This model of EMB design is displayed in Figure 1.1. 1 Specifically, I suggest that incumbents are more likely to implement significant reforms that enhance EMB autonomy and capacity when the costs of maintaining the status quo outweigh the benefits. I highlight three factors that influence the incumbents’ cost-benefit calculations: 1) the strength of domestic opposition relative to the incumbent and the ability of the opposition to credibly pressure the incumbent; 2) the regimes’ vulnerability to external pressure and the 1 All figures and tables are placed in the Appendix at the end of each chapter. 3 commitment of international actors to credibly pressure the incumbent; and, 3) the extent to which EMB reforms will influence the incumbents’ electoral prospects. Changes in EMB design are more likely to occur during critical junctures produced through exogenous or endogenous shocks to the political system. While there are many potential factors that make the possibility of EMB reform more likely, I focus on fraudulent elections involving extensive incumbent manipulation of electoral administration and administrative irregularities. Finally, significant changes in EMB design are less likely to occur when political elites consider the institution legitimate. It is during these times that EMBs experience lock-in effects and develop path dependent stability. Nevertheless, opportunities for more gradual change during periods of continuity exist because of the way political actors and EMB officials interpret, implement, and enforce existing rules. 1.2.2 Consequence of EMB Design and Performance To understand the consequences of EMB design and performance on election quality, I advance the historical institutionalism framework, and posit that there are conditions under which EMBs structure the incentives and strategic behavior of political actors during elections. EMBs matter for elites and citizens because they directly affect the costs to the incumbent and political elites for engaging in fraud and they also affect the extent of electoral administrative irregularities. The proposed theoretical model is highlighted in Figure 1.2. It shows that election quality measured through citizens’ and elites’ perspectives may be determined by (1) EMB design, (2) EMB performance, and (3) the overall political and institutional context. I define EMB design as the formal rules and procedures that define the functions, responsibility, and powers of the 4 2 electoral management body. EMB performance reflects the extent to which the formal rules are enforced and effectively implemented, the influence of informal rules, and the impact of the prevailing political context, including the rule of law, level of democracy, and institutionalization of other state and political institutions. While both EMB design and EMB performance are important predictors, I argue that in the context of emerging democracies EMB performance should more effectively explain variation in citizens and elites’ election quality attitudes and behavior. As a final step, I distinguish between EMB autonomy and capacity and apply this conceptual distinction to both EMB design (making the distinction between de jure autonomy and de jure capacity) and EMB performance (making the distinction between de facto autonomy and de facto capacity). When EMBs enjoy high levels of autonomy and capacity, the incumbent— and political elites more generally— find it more difficult to manipulate the EMB and engage in more general acts of electoral fraud. Moreover, EMB officials are better able to resolve the challenges of election administration and in so doing, reduce the magnitude of electoral administrative irregularities. All things being equal, I expect EMB autonomy and capacity to have positive impacts on the strategic interaction between elites and the overall quality of elections. In particular, political elites are more likely to play by the rules of the game (and eschew fraud and violence), participate in elections, and accept electoral defeat without crying foul. Similarly, I expect citizens to respond to variations in EMB performance in two primary ways. First, and 2 For instance, Section 153 of the 1999 Nigerian Constitution establishes the powers of the Independent National Electoral Commission in Nigeria. These powers include organizing all Federal and State elections, registering political parties, candidates and voters, and monitoring political party campaigns. Other rules regarding the political autonomy of the commission along with rules guiding the membership, appointment, and tenure of EMB executives are also established in the Constitution. 5 more directly, EMB performance can signal the incumbents’ commitment to making the playing field level and reducing the incidence of electoral irregularities. A second and more indirect effect is that citizens respond strategically to the interaction of elites. The extent to which elites play by the rules of the game, participate in elections, and accept electoral defeat have implications on how citizens view the quality of elections. 1.2.3 Linking Causes and Consequences Finally, I suggest that there is a cyclical relationship between the processes that determine the design and performance of EMBs and the consequences of EMB design and performance on election quality. From one perspective, the processes through which EMBs are designed has direct impacts on the distribution of political power and the extent to which elections are regarded as credible among the major players. For instance, citizens and political elites in countries where EMBs possess low levels of de jure autonomy and capacity are more likely to blame these institutional characteristics if elections are fraudulent and rigged. In turn, fraudulent elections could provide the impetus for domestic and international actors to change the existing EMB rules. On the other hand, free and fair elections may help legitimize the EMB and set the institution on a path dependent trajectory where institutional stability is established and further reforms are less likely to be needed (See Figure 1.3). 1.3 What is Electoral Administration? I situate the study of electoral administration within the general literature on electoral institutions (Mozaffar and Schedler 2002; Massicotte, Blais and Yoshinaka 2004). Although there have been various ways to classify electoral administration, I rely on the framework developed by Mozaffar and Schedler (2002). According to these authors, electoral governance is defined as “the wider set of activities that creates and maintains the broad institutional 6 framework in which voting and electoral competition take place.” There are three main levels which electoral governance occurs: 1) rule making; 2) rule application; and 3) rule adjudication. The various levels are displayed in Table 1.1. 3 Under the category of rule making, there are two broad categories of institutional rules that influence electoral governance. The first, and most widely studied, are the rules of electoral competition such as the electoral formula, district magnitude, electoral franchise, and the electoral timetable. Studies exploring the effects of rules associated with electoral competition 4 are too numerous to mention here. The second category includes rules that shape electoral administration (electoral governance). These rules encompass the design of electoral management bodies (electoral commissions), dispute resolution organizations (electoral courts), and processes such as voter registration, counting and tabulation of votes, and election observation. My research focuses specifically on the rules of electoral administration both from the perspective of rule making (EMB design) and also from the perspective of rule application (EMB performance). 1.3.1 Electoral Management Bodies Electoral management bodies (EMBs) are institutions vested with the responsibility to organize, supervise, and adjudicate on some or all of the stages of the election process (Mozaffar 2002; Lopez-Pintor 2000; Wall, for Democracy and Assistance. 2006). While the functions and 3 4 All tables and figures are placed in the Appendix at the end of the chapter. See the following for cross-national studies that examine the impact of electoral systems on citizens’ attitudes and behaviors: Birch 2008; Birch 2012; Norris 2004; Cox 1997, 1999; Powell 2002, 2006; Aarts and Thomassen 2008; Bowler and Donovan. 2012. See also specific cross-national studies in Africa: Lindberg, 2005; Lindberg 2006; Cho and Bratton 2006; Cho 2010, 2011; Vengroff and Ndiaye 1998. 7 responsibilities of EMBs vary across countries, some of the core functions and responsibility include • Registering voters and candidates; • Training electoral staff; • Conducting voting operations; • Counting and tabulating votes; and, • Announcing election results. Importantly, the design of EMBs are related to the rules of electoral administration, and EMBs are the institutions responsible for making sure that these rules are being implemented effectively and impartially during the election period. Policy-experts and scholars use “electoral management body” to formally classify institutions responsible for organizing elections. However, throughout the world there are other terms used to define these institutions, including electoral commissions (most developing countries such as Ghana and India), Boards of Elections (specific states in the USA such as South Carolina), Electoral Office (Jamaica), Secretary of State (specific states in the USA such as Michigan), and Ministry of Interior/Home Affairs (France, UK) (Wall, 2006). 1.4 Why Examine Electoral Administration? My research attempts to make important contributions to the existing literature on the role of political institutions in emerging democracies by concentrating on the part played by electoral management bodies in facilitating free, fair, and credible elections. There are many reasons that underscore my decision to study EMBs in emerging democracies. First, from a theoretical perspective, election administration is important to understanding the integrity of elections and the democratization process (Lopez-Pintor 2000: 13). In fact, we 8 might go so far as to regard EMBs as “instruments of democracy”(Powell 2004). This is especially the case in emerging democracies where there is less confidence in the legitimacy of other political and state institutions to not bias election administration in favor of the incumbent. As an institution, EMBs help ensure what Mozaffar and Schedler refer to as the “procedural legitimacy” and “substantive uncertainty” of elections by effectively managing various dimensions of the election process and holding political elites accountable to the rules of the electoral game (2002). The legitimacy of elections, one of the most crucial phases of the democratic process, depends on the performance of EMBs in conducting critical tasks such as the counting and tabulation of votes or the registration of voters. Furthermore, the burgeoning literature also associates EMB performance with the attitudes and behaviors of political elites and everyday citizens. In countries like Ghana and South Africa the perceived autonomy of EMBs has enhanced elite confidence in the integrity of the electoral system (Padmanabhan 2002; Debrah 2011). In countries such as Kenya (2007), Nigeria (2007), and Uganda (2006), lapses in EMB performance have had disastrous consequences for popular confidence in election quality, trust in the government, and satisfaction with democratic regimes (Adebayo and Omotola 2007; Leonard and Odhiambo-Owour 2009). Second, studying election administration also has important implications for public policy and democracy promotion. There is no gainsaying that effectively organizing elections is one of the most complicated, expensive, and logistically demanding exercises that a country will undertake. In fact, for many countries, elections involve the largest peacetime mobilization of a country’s financial and human resources (Lopez-Pintor and Fisher 2006). Meanwhile, the challenges of election organization increase exponentially in low and middle-income countries 9 that lack basic infrastructure (such as electricity so that counting of votes can be done in the night), adequate sources of educated poll workers, or sufficient financial resources to conduct technologically advanced processes such as biometric national voter registration. Because of the extensive scope and attendant challenges associated with electoral management, domestic and international policy-makers have devoted significant resources to improving the effectiveness and efficiency of electoral management bodies. For instance, many EMBs in Africa, such as Ghana, Kenya, and Zambia, have attempted to improve their voter registration capacity by developing new biometric voter registration systems (Evrensel 2010). Moreover, international donors such as the United States (through USAID), European Union, and the United Nations (through UNDP) have been instrumental in building the capacity of EMBs and promoting reforms in EMB autonomy. For instance, during the 2007 elections in Nigeria the UNDP through a donor basket fund contributed US $30 million to enhance INEC’s election preparations and to expand its capacity to promote stakeholder consultations in Nigeria (UNDP 2007). Finally, despite the clear theoretical and policy significance of EMBs, the role of these institutions has been under-theorized in the public policy and comparative politics literatures. Although there is a well-established literature on the determinants and consequences of rules that structure electoral competition (such as the level of proportionality, district magnitude, and electoral boundaries), less research has been carried out on the choice and consequences of rules shaping electoral administration (Mozaffar and Schedler 2002). A handful of recent studies examine factors influencing the choice of electoral administrative design (Mozaffar 2002; Gazibo 2006). However, more consideration should be given to factors that influence electoral 10 administration reform over time as well as the influence of international actors. I examine these two dimensions in this project. At the same time, scholars have also been researching the effects of election administration on political behavior (Elklit and Reynolds 2000; Birch 2008; Hartlyn, McCoy and Mustillo 2008). This emerging research is promising but there are some gaps that can be filled. One gap that I seek to fill in this dissertation is the effect of EMB capacity on elites and citizens’ electoral behavior. 1.5 Why Africa? Africa provides an ideal testing ground for examining the role of EMBs. First, there is significant variance in the institutional design and performance of EMBs (Elklit and Reynolds 2000; Mozaffar 2002). Figure 1.4 displays one dimension of EMB de jure institutional autonomy: the extent to which the executive branch of government is responsible for election management. Across 44 countries in Sub-Saharan Africa, the majority of EMBs (51 percent) are nominally independent, 16 percent are government administered, and 31 percent are a combination of government and independent models. I find similar variation across different dimensions of EMB capacity. Second, since the advent of multi-party democracy in the 1990s, many countries have undergone electoral administrative reforms, and the scope and frequency of these reforms vary (Lopez-Pintor 2000; Elklit and Reynolds 2005b; Gazibo 2006). Third, throughout Africa there is a wide range of political regimes (liberal democracies, electoral democracies, competitive authoritarian and authoritarian) and institutional contexts (van de Walle 2002). This variation helps in effectively testing the impact of EMB design and performance across a wide range on regime types. 1.6 Theoretical Approach 11 1.6.1 Formal and Informal Political Institutions My research examines the role of political institutions in emerging democratic settings. It joins a rich tradition of political science scholarship that examines the design of institutions and whether or not institutions “matter” for political elites and everyday citizens (Hammond and Butler 2003; Przeworski 2004). Following Douglas North, one of the pioneers of the institutional approach to political inquiry, I define institutions as the formal and informal rules of the game that provide both constraints and incentives for human interaction and behavior (1990). Specifically, highlighting the distinction between formal and informal rules is important for research on political institutions in emerging democracies. This is because the newly implemented democratic rules take a longer time to become institutionalized, while many institutional legacies provide important constraints on political behavior. As a result, our understanding of the choice and consequences of institutions should reflect the interaction of formal and informal rules and how they jointly influence political behavior. Numerous studies have demonstrated how informal rules can complement or undermine the influence of formal rules (Helmke and Levitsky 2006; Tsai 2006; Hale 2011). In particular, Helmke and Levitsky have theorized four different types of interactions between formal and informal rules (complementary, accommodating, competing, and substitutive) with varying effects on political outcomes (2006). For the purposes of this study, assessing the interaction between formal and informal rules is important for the following reasons. First, research on electoral management has highlighted how both formal and informal rules shape EMB performance. Some informal rules such as the development of inter-party political liaison committees in Ghana during the 1996 elections helped to enhance the legitimacy of Ghana’s Electoral Commission (EC) (Gyimah-Boadi 1999). By contrast, other scholars have noted how informal rules such as pervasive corruption and 12 clientelism can undermine the autonomy of EMBs (Eisenstadt 2004). Moreover, within the context of African neopatrimonial regimes, there are systemized informal rules, such as corruption, clientelism, and strong presidentialism, that influence the role and function of EMBs (Bratton and van de Walle 1997). 1.6.2 Institutional Design and Performance While in a perfect world, we would expect the performance of institutions to exactly reflect formal rules, various studies have pointed out that this is not always the case (Przeworski 2004; Weaver and Rockman 1993; Lijphart and Waisman 1996). According to Przeworski, institutions are endogenous and reflect “the conditions under which they emerge and endure” (2004:527). Because of this, I make a distinction between institutional design and institutional performance. I contend that institutional design is a function of the formal rules that define the institution’s powers and responsibilities. However, institutional performance is structured by the joint influence between formal and informal rules, the successful implementation and enforcement of these rules, and the dynamics of the wider politico-institutional context within 5 which the institution operates. As shown in Figure 1.2, there are three reasons why formal design may not translate into institutional performance. First, formal rules may not be effectively enforced due to actors or processes within and outside of the institution that block the successful implementation of institutions’ mandate (Huber and Shipan 2006:11). Second, informal rules may either “compete” with or “substitute” for formal rules (Helmke and Levitsky 2006). Finally, successful institutionalization requires a political and institutional context that provides a 5 The focus on institutional context is informed by Hammond’s work on bureaucratic autonomy, where autonomy can emerge in a political setting with multiple principals. According to him, “[t] he more principals there are in a political system, the more the bureaucrats are able to play these principals off against each other and thereby make decisions independent from them”(1996: 140). 13 conducive setting. This includes institutional safeguards for executive abuse of power as well as professional and capable bureaucracy, independent judicial institutions, and vibrant civic and political spaces. 1.6.3 Dimensions of Institutionalization: Autonomy and Capacity From a conceptual and analytical perspective, I consider autonomy and capacity as two interrelated but distinct dimensions of institutionalization. I define autonomy as an institution’s ability to make decisions independent of the control of the executive and other state and society actors. This conceptualization is informed by research on state autonomy (Geddes 1994; Huntington 1968; Skocpol 1985), bureaucratic autonomy (Huber and Shipan 2002; Hammond 1996; Carpenter 2001; Hammond and Knott 1996), and the autonomy of regulatory agencies such as the central banks (Moser 1999; Cukierman, Webb and Neyapti 1992). The literature utilizes a variety of concepts that are somewhat associated with autonomy, including independence, impartiality, neutrality, discretion, and political control. Chapter 2 provides greater detail on how my conceptualization of autonomy relates to these concepts. Capacity, on the other hand, refers to an institution’s ability to implement its legally mandated functions in an effective and efficient manner. This definition integrates theories of state capacity (Migdal 1988; Bratton 1994; Bratton and Chang 2006; Englebert 2000), bureaucratic capacity (Huber and McCarty 2004; Hall, Quin Monson and Patterson 2009; Ting 2005; Rauch and Evans 2000), and the capacity of other political institutions such as legislatures (Barkan 2009). Similar to autonomy, the existing literature has highlighted numerous dimensions of capacity: these will be addressed more fully in Chapter 2. The decision to descend the “ladder of abstraction” (Sartori 1970) and conceptually disentangle autonomy and capacity is based on the following logic. First, and most importantly, 14 scholars who examine the state, bureaucracies, legislatures, and other political institutions have made a similar conceptual distinction and have noted its theoretical import (Fukuyama 2013; 6 Bratton 1994). The second reason is more empirical. Existing research has indicated that at a given time an institution may have disproportionate levels of autonomy and capacity. The possibility that institutions may have disproportionate levels of autonomy and capacity has theoretical significance. From the perspective of institutional design, it raises the prospect that factors and processes, which influence autonomy, may not always be the same as those which influence capacity. Similarly, the consequences of capacity on elite behavior may be different from those of institutional autonomy. Finally distinguishing between autonomy and capacity is important because it is likely that autonomy has an influence on capacity and vice-a-versa (Geddes 1994). I apply these insights to the study of EMBs. Table 1.2 displays the levels of EMB autonomy and capacity in Ghana, Nigeria, South Africa, and Zimbabwe over time. Table 1.2 provides empirical justification for the conceptual distinction between EMB autonomy and capacity. While EMBs in some countries may have similar levels of autonomy and capacity (e.g, Ghana 1992: low capacity and low autonomy), EMBs in other countries have disproportionate levels of autonomy and capacity (e.g, Zimbabwe 2000-2008: low autonomy and high capacity). Furthermore, the levels of EMB autonomy and capacity can change over time. For example, South Africa that moves from having high autonomy and low capacity during the 1994 elections to having high levels of both autonomy and capacity. Additionally, the variation in EMB autonomy and capacity across the four countries highlighted in Table 1.2 provides justification 6 The theoretical basis for a distinction between EMB autonomy and capacity is informed by Bratton’s conceptual separation of political autonomy and political capacity in which “autonomy refers to the process in which actors set goals for organizations whereas capacity signifies the means of goal realization” (1994: 236). 15 for the selection of these four cases in the qualitative comparison in Chapter 3 and qualitative interviews that inform various chapters in the dissertation. These country-cases were selected in order to maximize the possible variation in EMB autonomy and EMB capacity. 7 1.6.4 Historical Institutionalism As indicated previously, I utilize historical institutionalism as a unifying framework to explain the variation in EMB design as well as the consequences of EMB performance on how citizens and elites perceive election quality. There are three reasons why historical institutionalism informs my research on the causes of EMB design. First, historical institutionalism provides a dynamic framework that highlights clear mechanisms for explaining periods of institutional stability and points of innovation through a focus on 1) path dependent trajectories (institutional and cultural legacies), 2) distributional feedback mechanisms that reinforce institutional stability (emphasis on the strategic conflict between political actors), and 3) and the emergence of critical junctures (exogenous and endogenous mechanisms of change). Second, the framework focuses on the interplay between agential and macro-structural explanations of change and fully embraces the complexity of the political world by highlighting the influence of non-political factors (Bratton 8 and van de Walle 1997). Finally, historical institutionalism has been used to explain various 7 8 Details about the case selection criteria are discussed in Chapter 2. For instance, Bratton and Van de Walle (1997) use historical institutionalism to explain democratic transition outcomes in Africa, while Boone (2003) uses a variant of institutionalism to explain variations in the design of local level institutions in rural Africa. 16 9 accounts of institutional change in comparative politics and figures prominently in the emerging 10 research on electoral administrative reform (Mozaffar 2002; Gazibo 2006). There are three dominant approaches to evaluate the effect of institutions on political behavior: rational choice, sociological and historical institutionalism (Thelen 1999; Hall and Taylor 1996). 11 Of the three analytical approaches, historical institutionalism is most useful for my research. Historical institutionalism represents the ideal synthesis of the instrumental (rational choice) and cultural approaches (sociological) of explaining institutional influence (Hall and Taylor 1996). Borrowing from rational choice, historical institutionalism incorporates the notion that institutions structure the strategic interaction of utility maximizing political actors by providing information on the strategies that other actors employ and the sanctions for shirking. However, unlike their rationalist counterparts, historical institutionalists believe that institutions not only shape strategies but also the preferences and goals of individuals. Moreover, since there are contexts in which individual action is not driven by rational-instrumentality, historical institutionalism’s reliance on non-instrumental explanations of behavior rooted in the influence of cultural and social norms increases its theoretical applicability. 9 12 Steinmo, Thelen and Longstreth (1992) provide a general overview. For applications in Africa see Bratton and van de Walle 1997 and Boone (2003). 10 Historical institutionalism is not the only framework for understanding institutional origin and development; another ascendant framework is Elinor Ostrom’s (1990) institutional analysis and design (IAD). 11 For comprehensive surveys of the “new” institutionalism literature see the following: Carey (2000); Engerman and Sokoloff (2008); Olsen and March (1984). 12 Scholars have often criticized historical institutionalism for its limited attention to theory building and clear specification of causal mechanisms between institutions and political behavior (Hall and Taylor 1996). According to Thelen, such criticisms are less applicable to more recent works within the historical institutionalists tradition because they often contain mid-level theories. Thelen also argues that any deficiencies in theory building are outweighed by use of rigorous empirical testing (1999:72-73). 17 According to historical institutionalists, institutions such as EMBs provide vital information on the expected behavior of other actors as well as the effectiveness of enforcement mechanisms to ensure that actors play by the rules. Another important feature of institutions is that they provide coordinating mechanisms that help reduce the transaction costs of cooperation. These informational and coordinating features are particularly important for our assessment of EMBs in emerging democratic contexts. In regimes where there is a legacy of overbearing authoritarian rule and the current institutional context fails to provide sufficient checks against executive manipulation, EMBs can help provide institutionalized democratic uncertainty; that is, the outcomes of elections are not pre-determined. My research examines the role of electoral management specifically in the context of emerging democracies. As a result, it relies heavily on the democratization literature to understand the dynamic processes and institutional structures that inform the preferences and strategic interaction of political actors. One important dimension of democracy is the quality of elections and how free and fair elections can enhance the prospects of democratic consolidation (Diamond 1994). Democratic theorists have conceptualized election quality in myriad ways (Schedler 2002b; Dahl 1971; Svensson and Elklit 1997). This dissertation adopts the conceptualization of Elklit and Reynolds (2000; 2005) in which election quality represents “the extent to which the entire electoral process is seen as legitimate and binding by political actors” (2000:1). High quality elections require freedom and fairness, hence the common terminology “free and fair” elections. 13 Freedom connotes candidates’ ability to run for elected office and engage in 13 For more on the conceptualization of election quality see: Boda (2005); Diamond (2002); Goodwin-Gill (1994); and Massicotte (2005). 18 campaigning, plus the voters’ ability to choose a preferred outcome without constraint or coercion. Fairness signifies the impartial and consistent application of rules (international and domestic) in ways that do not adversely bias some political competitors relative to others. The fairness dimension is often evidenced through impartial and effective conduct of electoral-related institutions such as the electoral commission, security forces, media and courts, and the absence of systematic electoral fraud and manipulation. More recent studies have also associated election quality with concepts of electoral integrity and electoral malpractice. These more recent conceptualizations encompass similar processes through which political actors develop their election quality opinions (Birch 2011; Norris 2011). 14 1.7 Research Methods I probe the causes and consequences of electoral administration in the context of Africa using a mixed-methods approach. I embed a four-country comparative analysis of EMBs in Ghana, Nigeria, South Africa, and Zimbabwe within a rigorous cross-national and cross-time analysis of more than 20 African countries. Importantly, this multi-method approach incorporates data from a variety of sources, including original cross-national data on EMB design, existing cross-national data on EMB performance, cross-national survey data, 88 original elite interviews that I personally conducted in Ghana, Nigeria, South Africa and Zimbabwe, and participant observation of elections in Nigeria and Ghana. The purpose of integrating multiple methods through what Sydney Tarrow refers to as “triangulation” is aimed at increasing the amount of information I utilize to evaluate the main inferences within the dissertation (Tarrow 14 According to Norris, electoral integrity denotes “shared international principles, values, and standards of democratic elections which apply universally to all countries and which should be reflected at all stages during the electoral cycle, including the pre-electoral period, the campaign, polling day and its aftermath.” Electoral malpractice, on the other hand, refers to attempts to violate electoral integrity (2011). 19 1995). Moreover, the integration of qualitative and quantitative methods also helps overcome some of the common challenges, such as problems of bias and spurious findings, that emerge when only one type of method is pursued (Lieberman 2005). 15 1.8 Research Contributions From a theoretical perspective, the research advances the existing literature on political institutions, electoral governance, and electoral reform by proposing a framework to explain both the causes and consequences of election administrative reform. Additionally, the research makes a conceptual contribution by distinguishing between two dimensions of EMB performance: institutional capacity and autonomy. I explore both the determinants and consequences of the two dimensions within countries over time and across countries. There are several empirical contributions. First, the research represents the first study to rigorously examine both the causes and consequences of EMB design and performance on African elites and citizens using cross-national data. By doing so, I expand upon existing crossnational qualitative (Gazibo 2006) and quantitative (Mozaffar 2002) research that explores the factors influencing EMB design in Africa. I also fill the gaps within the existing cross-national literature that examines the consequences of EMB design and performance (Birch 2008; Hartlyn, McCoy and Mustillo 2008; Rosas 2010). From a policy perspective, this research rigorously evaluates conditions that make EMB reform most likely as well as the conditions under which EMBs can most effectively reduce the incidence of procedural irregularities and deter incentives to commit fraud. This information may help enhance the strategies and approaches that organizations in Africa and the developing world 15 Chapter 2 provides a detailed description of the research methods used in the dissertation. 20 pursue when seeking election administrative reform. Furthermore, my research has the potential to advance policy practitioners’ understanding of the different ways in which EMB institutional capacity and autonomy shape political confidence in the EMB and perceptions of election quality in a wide range of political settings. 1.9 Roadmap of Subsequent Chapters Chapter 2 outlines the various methods used in the dissertation to examine the causes and consequences of EMB design and performance in Africa. I begin by surveying different approaches for conceptualizing and measuring institutional autonomy and capacity in the literature and demonstrate how recent studies on election administration have not adequately integrated electoral administrative capacity. Following this, I introduce the dataset on election administrative reform across 20 countries in Africa from 1990 to 2008 and detail the methods for data collection. I also highlight the methods for conducting the qualitative small-n comparative study in Ghana, Nigeria, South Africa, and Zimbabwe, including the selection of research sites and the sampling methods for elite interviews. Chapter 3 examines a theoretically intriguing puzzle: why some incumbent African leaders consent to reforms that will enhance the autonomy or capacity of election administration when doing so may undermine the leaders’ ability to manipulate elections and secure re-election. I survey different approaches to explaining institutional change in the comparative politics and public policy literature and argue for the utilization of historical institutionalism as a framework. I then propose a framework that links election administrative reform to the strategic interactions of domestic and international actors along with the path dependency of institutional legacy during different stages of democracy. I use this framework to conduct a comparative analysis of election administrative reform in Ghana, Nigeria, South Africa and Zimbabwe from 1990-2008. 21 Using data collected in elite interviews and archival research, I engage in detailed comparative tracing of the electoral reform process by identifying the actors (and preferences) and political contexts that influence the success or failure of reform. The findings point to the important role that a coalition of opposition, international donors, and civil society actors has in compelling incumbents to enact electoral administrative reform. However, the scope of the reform process is also shaped by other factors such as path dependence and the degree of irregularities in previous elections. In Chapters 4, 5, 6 and 7 I shift the focus to investigating the consequences of EMB design and performance for the quality of elections in Africa. Chapter 4 assesses the impact of EMB design on elite and citizens’ perspectives of election quality. I distinguish between two main dimensions of EMB —de jure autonomy and de jure capacity— and find that neither dimension has an effect on whether citizens or elites consider elections free and fair. I associate these findings with the deliberate attempts of incumbents to block the implementation of electoral rules that will impinge their ability to rig elections. Chapters 5 and 6 explore the impact of EMB performance on the incentives and behaviors of elites and citizens respectively. In Chapter 5, I demonstrate how EMB performance can directly reduce the mechanisms by which the incumbents and other political actors can manipulate election administration and other dimensions of election processes. As a consequence, elites may be more willing to participate in elections, to adhere to electoral rules, and to accept unfavorable of results. I identify four indicators of elite perspectives on election quality: opposition willingness to boycott elections, overall election quality evaluations, losers’ acceptance of results, and electoral turnover. Then I statistically evaluate the relationship between these four indicators of election quality and our measures of EMB performance across a 22 45-country sample. The regression results indicate a strong and positive relationship between election administrative performance and the indicators of election quality. This provides substantial support for the importance of EMB performance for election quality in Africa. Chapter 6 considers whether citizens’ attitudes towards elections are influenced by the performance of election administration in their respective countries. I start by surveying the existing literature on the determinants of citizens’ election quality perceptions and highlight various ways through which election administration shapes citizens’ perspectives on election quality and their willingness to participate in elections. I identify one indicator of citizens’ perspectives of election quality: perceptions of the freeness and fairness of the electoral context from Afrobarometer surveys between 1999 and 2008. Using a multi-level regression analysis of 20 sub-Saharan countries, I explore the relationship between popular election quality perceptions and election administrative performance. The findings indicate a positive relationship between election administrative performance and citizens’ election perceptions as well as their willingness to participate in elections. Chapter 7 expands the analysis on how election administration shapes citizens’ opinions of election quality through a more focused examination of citizens’ attitudes in Nigeria between the 1999 and 2011 elections. The Nigerian case provides an ideal context because the country experienced election administrative reforms in 1999, 2007, and 2010. Furthermore, in both 2007 and 2011 there were a series of surveys that directly gauged Nigerians’ knowledge of electoral reform, evaluations of electoral administrative autonomy and capacity, and their views on election quality. Using the data from these surveys, I conduct a series of empirical analyses on the relationship between citizens’ evaluations of EMB autonomy and capacity and their perceptions of the quality of elections. I find that Nigerians are capable of distinguishing 23 between the autonomy and capacity of election administration and use these evaluations along with other factors to evaluate the quality of the elections. Importantly, the results also show that Nigerians who negatively evaluated the autonomy and capacity of INEC were also less willing to consider elections free and fair. Chapter 8 summarizes the main findings of the dissertation and examines the main theoretical and policy implications. Furthermore, I detail some of the main theoretical and empirical limitations and highlight potential areas of future research. 24 APPENDIX 25 Table 1.1: Three Levels of Electoral Governance Levels 1. Rule Making: Choosing and defining the basic rules of the electoral game Elements a) Rules of Electoral Competition: Formula District magnitude District boundaries Assembly size Electoral time table Franchise b) Rules of Electoral Governance: Voter registration Party and candidate registration Campaign financing and regulation Election observation Ballot design Polling stations Voting, counting, and tabulating Election management bodies Dispute settlement authorities 2. Rule Application: Organizing the electoral game. Registration of voters, candidates, parties Registration of election observers Voter education Electoral organization Voting, counting, and reporting 3. Rule Adjudication: Certifying election results and resolving disputes. Admission of complaints Processing of cases Publication and implementation of rulings Source: Mozaffar and Schedler (2002) 26 Table 1.2: Distribution of EMB Autonomy and Capacity EMB Autonomy Moderate/ High Low EMB Capacity Low Moderate/High Ghana (1996) South Africa (1999 -2009) South Africa (1994) Ghana (2000-2008) Nigeria (2011) Ghana (1992) Zimbabwe (2000-2008) Nigeria (1999, 2003, 2007) 27 Table 1.3: Distribution of EMB Models in Africa Government Togo Cote d'Ivoire Guinea Djibouti Mauritania Rwanda Seychelles Mixed Burkina Faso Burundi Cameroon Cape Verde Central African Republic Chad Comoros Republic of Congo Equatorial Guinea Gabon Madagascar Mali Senegal Togo Independent Angola Benin Botswana Congo (Dem Rep) Gambia Ghana Guinea (Conakry) Guinea-Bissau Kenya Lesotho Liberia Malawi Mauritius Mozambique Namibia Nigeria Nigeria Niger Sierra Leone South Africa Tanzania Uganda Zambia Zimbabwe N=7 (16%) N=14 (31%) N=24 (53%) Source: Electoral management design: the International IDEA handbook (2006) 28 Figure 1.1: Theoretical Model: Causes of EMB Design and Performance CRITICAL JUNCTURES INTERNATIONAL COMMUNITY EMB DE JURE AUTONOMY POWER ASYMMETRY Costs of Status Quo EMB DESIGN EMB DE JURE CAPACITY DOMESTIC OPPOSITION POLITICAL & INSTITUTIONAL CONTEXT 29 Figure 1.2: Theoretical Model: Consequences of EMB Design and Performance Informal Rules Elite Attitudes & Behavior Formal Rules EMB Performance (De Facto Autonomy & Capacity) EMB Design (De Jure Autonomy & Capacity) Citizen Attitudes Political & Institutional Context 30 Figure 1.3: Cycle of EMB Reform and Election Quality EMB DESIGN DEMAND FOR REFORMS EMB PERFORMANCE ELECTION QUALITY 31 CHAPTER 2 DATA AND METHODS 2.1 Introduction This chapter outlines the various methods used in the dissertation. I begin in Section 1 by briefly surveying the existing data on EMB design and performance. Section 2 describes the rationale for constructing an original dataset on EMB design in Africa and details the methods for data collection. Section 3 outlines the other sources of data used in the large-n analysis, including EMB performance and expert and survey measures of election quality. Section 4 explains the methods used in the four-country comparative study, including the selection of research sites and the sampling methods for elite interviews. 2.2 Existing Data Sources on EMB Design and Performance 2.2.1 Data on EMB Design Currently there is only a handful of studies that examine the role of electoral management bodies using cross-national data. Of these studies, most have focused on collecting data on the design of EMBs, while in the last two years two innovative datasets have been established to examine the performance of EMBs. I will begin by looking at the main data sources on EMB design. Allan Wall and his colleagues developed one the most comprehensive sources of data on 1 EMB design (2006). The dataset includes cross-national information on institutional autonomy, professional autonomy (appointment process and tenure of EMB commissioners), and partisan 1 Institutional autonomy measures the extent to which the executive branch of government is responsible for the administration of elections. EMBs can be government-run, mixed (both government and independent) or independently administered. 32 autonomy (whether the EMB is based on partisan or expert membership) for all countries in the world in 2006. One shortcoming of the dataset is that it does not highlight the overtime changes in these dimensions in EMB design. Jonathan Hartlyn and his coauthors (2008) developed a similar dataset on the design of election administration in Latin America that includes indicators of EMB professional autonomy (appointment process and tenure of EMB commissioners) and EMB partisan autonomy (extent of political party involvement in EMB membership). What distinguishes their data is its focus on changes in the autonomy of EMBs in 19 Latin American countries over a 20-year period. Subsequent research on EMBs in Latin America by Guillermo Rosas (2010) utilized similar coding criteria for partisan and professional autonomy. To date, very few empirical studies have developed indicators of EMB design in Africa using cross-national data. Shaheen Mozaffar (2002), in his study on the variation of EMB design in Africa, focuses on one dimension: institutional autonomy. Similar to Wall et al., Mozaffar’s dataset is limited to one dimension of EMB design, institutional autonomy, and only examines the variation using cross-country data at one time point. There are very few existing data sources on EMB de jure capacity. Again, this reflects the literature’s disproportionate focus on the autonomy of EMBs. There have been studies that include data on specific dimensions of EMB de jure capacity such as the rules and procedures shaping voter registration (Schaffer 2008; Evrensel 2010), voter operations (Tomz and Houweling 2003; Alvarez, Hall and Llewellyn 2008), and counting and tabulation of results (Schaffer 2008). Nonetheless there are no studies that have systematically developed measures for multiple dimensions of EMB capacity based on cross-national and cross-time data. To overcome this, I rely on frameworks developed by policy-makers and election observation 33 organizations to evaluate the capacity of EMBs during different phases of the election process (Elklit and Reynolds 2005a). 2.2.2 Data on EMB Performance The study of EMB performance has seen the most recent improvements in the availability of cross-time and cross-sectional data. The first database was developed by Judith Kelley to gauge the quality of elections across sovereign states between 1945 and 2004 (Kelley and Kolev 2010; Kelley 2012). Kelley codes these data from the U.S. State Department Human Rights Reports for the countries under investigation. Among the many measures of election quality, Kelley includes two indicators of the extent of administrative problems experienced in the preelection and election day stages of the electoral process. What is significant about these two indicators is that they reflect important dimensions of the actual or de facto EMB autonomy and de facto capacity. For example, the measures reflect key aspects of EMB autonomy including the impartiality of the appointment process in practice and how evenly the EMB has been enforcing the electoral law. With respect to capacity-related attributes, the measure captures logistical problems at voting locations, problems associated with election day workers, and deficiencies with the voter’s register. Like many cross-national indicators of institutional characteristics, Kelley’s measures of electoral administrative problems have important shortcomings. First, the indicator does not allow researchers to disaggregate the autonomy and capacity dimensions. Additionally, some may criticize the measure for focusing more on the capacity-related dimensions of the EMB performance. Despite these shortcomings, I am confident that the indicator provides the most accurate and reliable gauge of EMB performance that allows for meaningful cross-time and 34 cross-national comparisons. I utilize this measure as the main indicators for the performance of EMBs in Chapters 5 and 6. Birch (2011) has developed a similar cross-national dataset on electoral malpractice that includes indicators of the independence/impartiality of EMB officials (indicator of autonomy), the quality of voter registration process, and the adequacy of polling arrangements, voting processes, and counting and tabulation activities (indicators of capacity). Unfortunately, however, the dataset is limited in the temporal scope (1995-2007) and the number of cases in Africa (19) and was not utilized in this research project. 2.3 Constructing a Dataset on EMB Design in Africa To address the limitations in available data on EMB design in Africa, I created an original dataset that includes several indicators of EMB de jure autonomy and de jure capacity across 20 African countries between 1990 and 2008. Because of the availability of reliable data 2 on EMB performance (Kelley 2012), I focused only on EMB design. These 20 countries were deliberately selected to correspond with the 20 countries included in Round 4 of the 3 Afrobarometer public opinion survey. Despite the purposive selection of cases, the countries included within the sample exhibit significant cross-time and cross-country variation in EMB de jure autonomy and capacity. It is important to note that the countries included in the sample are not fully representative of African countries; these countries tend to be more democratic than the average African regime (Cho 2010). 2 The twenty countries include Benin, Botswana, Burkina Faso, Cape Verde, Ghana, Kenya, Liberia Lesotho, Madagascar, Malawi, Mali, Mozambique, Namibia, Nigeria, Senegal, South Africa, Tanzania, Uganda, Zambia, and Zimbabwe. 3 This decision was taken so as to effectively carry out an analysis of the relationship between EMB design and popular election quality perceptions using Afrobarometer data. 35 The unit of analysis is the election period that begins after election in time t and ends on the day of election in time t + 1. Based on this the dataset includes 95 election periods. The data are coded from a wide range of sources including international legal instruments, country constitutions, electoral codes, and EMB regulations and procedures. Additionally, I incorporate analysis of EMB de jure autonomy and de jure capacity found in reports by international and domestic observer organizations, scholarly accounts of the elections, as well as reports from domestic and international NGOs and reports in popular media. 2.3.1 EMB Autonomy The coding for the indicators of de jure autonomy is derived from the existing data on EMB autonomy (Wall, for Democracy and Assistance. 2006; Hartlyn, McCoy and Mustillo 2008). I focus on four main dimensions of de jure autonomy that are further disaggregated into individual measures: 1. Institutional autonomy a. Institutional model: measures the extent to which the executive branch of government is responsible for the administration of elections. EMBs are government-run, mixed (both government and independent), or independently administered. b. Legal autonomy: indicates whether the specific model of election management is stipulated in the constitution or electoral law. I expect that when election management is constitutionally defined, political elites find it more difficult to alter key aspects of the electoral administrative framework through legislative reform. 2. Professional autonomy 36 a. Appointment process of EMB commissioner b. Tenure length of EMB commissioner c. Removal process of EMB commissioner 3. Partisan autonomy a. Extent to which EMB is comprised of partisan or expert members 4. Financial autonomy a. Degree of EMB financial control over operational budget 2.3.2 EMB Capacity The coding for the indicators of de jure capacity is partly derived from a framework for assessing EMB capacity developed by Elklit and Reynolds (2005a) and a combination of other sources (Fall et al 2011; Birch 2011). In the same way, I focus on seven main dimensions of de jure capacity that are further disaggregated into individual measures: 1. Professional Capacity a. Whether EMB commissioners need to meet professional qualifications for appointment 2. Administrative Capacity a. Whether the EMB secretariat is permanent or established for each election period 3. Voter Registration Capacity a. Extent to which the voter registration process allows for the accurate collection of information on voters 4. Voting Operations Capacity a. Process for voting 37 b. Process for counting and tabulating results c. Process of announcing results 5. Conflict Resolution and Stakeholder Cooperation Capacity a. Whether the EMB has mechanisms to facilitate cooperation among political parties 6. Legislative Capacity a. Whether the EMB has the power to propose electoral administrative legislation b. Whether the EMB has the power to pass electoral administrative regulations 7. Sanctioning Capacity a. Whether the EMB has the capacity to sanction political actors who violate election regulations? 2.3.3 EMB Reform A second purpose of developing the dataset of EMB design in Africa was to track changes in constitutional, statutory, or procedural rules affecting our four main dimensions of EMB de jure autonomy or the seven main indicators of EMB de jure capacity. These changes are considered electoral administrative reforms. Of the 95 election periods included in the sample, constitutional reforms occurred in 36 percent of election periods (34 cases), legislative reforms occurred 78 percent of election periods (74 election periods), while procedural changes occurred in virtually every election period. 2.4 Additional Large-n Data Sources In Chapters 4 through 7, I examine the consequences of EMB design and performance on the quality of elections. To gauge elite electoral attitudes and behaviors, I use aggregate indicators of overall election quality, opposition participation in elections, losers’ acceptance of 38 results, and electoral turnover. These four indicators are obtained from Staffan Lindberg’s 4 Dataset on Elections in Africa (2006a). The dataset includes information on the aforementioned indicators across 48 African countries between 1989 and 2006. The main indicator for citizens’ perceptions of election quality was obtained from a single question on the Afrobarometer public 5 opinion survey. Specifically, I relied on data from Rounds 1-4 of the Afrobarometer that 6 includes information on elections in a total 20 countries between 1999 and 2008. For the analyses of citizens’ election quality perceptions in Nigeria and their evaluations of EMB performance, I incorporate data from various public opinion surveys including the Afrobarometer, 7 International Foundation of Electoral Systems (IFES), 9 8 International 10 Republican Institute (IRI), and NOI Polls. 2.5 Research Methods: Small-n Analysis I supplement the large-n statistical analyses with a qualitative cross-national comparative analysis of the causes and consequences of EMB design and performance in Ghana, Nigeria, South Africa, and Zimbabwe. 4 For more information on coding of the measures please see Appendix A1. Also see http://www.clas.ufl.edu/users/lindberg/ for more information on the complete data set. 5 The Afrobarometer asks: “On the whole, how would you rate the freeness and fairness of the last national election?” Response options include “Not free and fair,” “ Free and fair, with major problems,” “Free and fair, but with minor problems,” and “Completely free and fair.” 6 For more information on the Afrobarometer see: http://www.afrobarometer.org/ 7 8 Afrobarometer conducted surveys in Nigeria in 1999, 2003, 2007, 2008, and 2012. IFES conducted two election surveys in February and May 2007. For more information see: ifes.org 9 IRI conducted two election surveys in 2010 and 2011. For more information see: http://www.iri.org. 10 NOI Polls has conducted several polls in Nigeria before and after the 2011 elections. For more information see: http://www.noi-polls.net/. 39 2.5.1 Case Selection These four countries were selected using two main criteria. First, and most importantly, the cases were selected so as to effectively conduct the comparative study on the causes of EMB design (reform) in Chapter 3. In particular, I wanted to include cases with significant cross-time and cross-case variation of the main dependent variables: EMB de jure autonomy and de jure 11 capacity. As shown in the Table 2.1, the four countries vary with respect to the levels of both EMB de jure autonomy and capacity both over different election periods in a specific country and across countries. 12 Second, I followed a most similar systems research design, whereby two or three of the countries selected shared similar values on some of the factors highlighted in the literature as affecting the design of EMBs (Seawright and Gerring 2008). For instance, Mozaffar (2002) identifies colonial legacy as an important determinant of EMB design. Hence, I controlled for the influence of colonial legacy by ensuring that the four countries were former British colonies. Even more, the countries in the sample also share similarities with respect to post- colonial regime type: Ghana and Nigeria experienced periods of military rule, while Zimbabwe and South Africa had white-minority oligarchies (Bratton and van de Walle 1997). Also pairs of countries share similarities with respect to key population size, economic development, and integration into the global economy. 2.5.2 Semi-structured Interviews 11 12 All figures and tables are placed in the Appendix at the end of each chapter. This will be discussed in greater detail in Chapter 3. 40 I also carried out a total of 88 semi-structured interviews in each of the four countries between 2010 and 2012. 13 The semi-structured interview format is ideal for this type of research as it allows for the collection of data that is comparable across subjects, but it also gives subjects the freedom to divulge information that reflects their individual experience. Importantly, the data gathered from the semi-structured interviews provides crucial information on the factors that have most recently influenced the design of EMBs as well as the various ways that EMBs influence popular and elite attitudes towards election quality. 14 The sample universe includes elite election stakeholders, who are further organized into five categories: 1) executive electoral administrators and civil servants; 2) executive political party officials, legislators, members of government; 3) leaders of domestic and international NGOs; 4) journalists; and 5) academics. I created a sampling frame that includes a list of potential election stakeholders for each research site. The database was compiled by incorporating existing databases established by 15 domestic and international NGOs operating in each country as well as references to election stakeholders made in the popular media. I then organized the database according to the five main 13 Because of Institutional Review Board (IRB) protocols, the specific names of the interviewees are not mentioned directly. With the exception of a few interviewees who specifically gave permission for their names to be disclosed, all other interviewees are referred to by their specific category (for instance, “political party official: member of parliament”). In developing the research design for IRB approval it was thought that potential subjects would be more willing to divulge sensitive information if their names were kept confidential. This is of primary concern in research sites such as Zimbabwe where members of the opposition and civil society are often detained for criticizing the government. In lieu of the names of the interviewees I provide a Table that outlines the number of interviews conducted in each research site (See Table 2.2). 14 A sample interview script is including in the Appendix A2. 15 For instance in Ghana, I gained access to the database on election stakeholders from the Carter Center and National Democratic Institute, a leading research and policy institute in Ghana. 41 categories of election stakeholders. Potential subjects were selected from each category by using a sampling procedure that considered the subject’s position, engagement in elections, and when possible their gender, ethnic, political and regional background. The specific procedure was used to ensure the representativeness of the sample of election stakeholders. For each country, I selected four to five individuals from each category of stakeholders. While the sampling procedure worked well in the four research sites, there were some stakeholder groups for which it was difficult to get access to potential stakeholders. In these cases, I utilized a snowball sampling technique in which interviewees were asked to suggest other potential election stakeholders. For instance, this technique was used to gain access to senior politicians in Zimbabwe. 2.5.3 Participant Observation In addition to conducting elite interviews, I had the opportunity to immerse myself in electoral management bodies and engage in the “soaking and poking” that Putnam identifies as a crucial component of research on institutional performance (1993). I had the opportunity to observe elections in Ghana (2008) and Nigeria (2011) with two international election observation organizations: the Carter Center and the National Democratic Institute. Through these two observation missions, I gained access to a wide cross section of elites, including leading members of political parties and election management officials. I also gained unhindered access to various stages of the electoral process including voter registration and the often contentious counting and tabulation process. In particular, by serving as a long-term observer during the 2011 Nigerian elections, I was responsible for meeting with numerous stakeholders, and reporting on various phases of the election process in six states in the South West geo-political zone (including Ekiti, Lagos, Ondo, Ogun, Osun and Oyo states) over a five-month period. 42 APPENDIX 43 Table 2.1: Distribution of EMB Autonomy and Capacity by Election Periods EMB Autonomy Moderate/ High Low EMB Capacity Low Moderate/High Ghana (1996) South Africa (1999 -2009) South Africa (1994) Ghana (2000-2008) Nigeria (2011) Ghana (1992) Zimbabwe (2000-2008) Nigeria (1999, 2003, 2007) 44 Table 2.2: Research Sites and Number of Elite Interviews Country Research Locations Number of Interviews Ghana Accra Region Kumasi Region 14 5 Nigeria Lagos State Ogun State Abuja FCT 10 2 15 Cape Town Johannesburg/Pretoria 12 9 Harare 21 88 South Africa Zimbabwe Total 45 Interview Question Script for Semi-Structured Interviews [Zimbabwe] INTRODUCTORY SCRIPT Thank you for agreeing to be interviewed for this research study. The study, which is part of my doctoral dissertation research at Michigan State University, investigates the causes and consequences of electoral administrative reform. The questions in this interview will focus on 1) the factors that influence governments to improve electoral administration, 2) the performance of electoral management bodies (EMBs) in recent elections; and 3) the effect of EMB performance on the quality of elections. You may choose not to answer a particular question if you feel uncomfortable about answering at any time during the interview. SECTION 1: Respondent Information 1. Age 2. Gender 3. Organizational Affiliation 4. Which [Zimbabwean] language do you speak most often? SECTION 2: Legal and Constitutional Framework 1. What is the formal process for appointing the commissioners of Zimbabwe Electoral Commission (ZEC)? a. How does it actually work in practice? 2. What is the process for appointing regional directors of elections? a. How does it actually work in practice? 3. How does the ZEC formally finance its operations? a. How does it actually work in practice? SECTION 3: Confidence in EMB 1. How much confidence do you have in the Zimbabwe Electoral Commission (ZEC)? a. Can you tell me why you have so [much/little] confidence in ZEC? 2. Can you identify the chairman of the Zimbabwe Electoral Commission (ZEC)? 3. How much confidence do you have in the chairman of the Zimbabwe Electoral Commission (ZEC)? a. Can you tell me why you have so [much/little] confidence in ZEC? 4. Can you identify the Regional Director of Elections in of your region? 5. How much confidence do you have in the Regional Director of Elections of your region? a. Can you tell me why you have so [much/little] confidence in the Regional Director of Elections in your region? SECTION 4: Performance of EMB in most recent election 1. How would you evaluate the Zimbabwe Electoral Commission’s (ZEC) performance in conducting the March 2008 national elections? 2. Please tell me how satisfied or dissatisfied you were with Zimbabwe Electoral Commission’s (ZEC) performance the following aspects of the December 2008 national elections. Can you explain why you are satisfied or dissatisfied (verbatim)? 46 a. Voter Registration Process b. Information on where and when to vote b. Competency of the polling station staff c. Adequacy of facilities and equipment at the polling station d. Counting process at polling station e. Tabulation process at ZEC centers and announcement of results 3. In your opinion did the ZEC have the necessary financial resources to effectively organize the 2008 national elections? 4. Thinking about the performance of the Zimbabwe Electoral Commission (ZEC) in the March 2008 national elections, tell me whether you disagree or agree with the following statements? a. Zimbabwe Electoral Commission (ZEC) performed its duty as a neutral body guided in its work only by the law. b. Zimbabwe Electoral Commission (ZEC) equally rep- resented the interests of different political parties b. During the various stages of the election process how often were you in contact with members of the ZEC? c. How well or badly would you say the Zimbabwe Electoral Commission (ZEC) handled the dissemination of information about the electoral process? d. How well or badly would you say the Zimbabwe Electoral Commission (ZEC) worked with other stakeholders [political parties, media, civil society and security officials] to ensure peaceful and credible elections? SECTION 5: Electoral Administrative Policy 1. How effective do you think the Zimbabwean election legal framework was in helping to make the 2008 national elections both free and fair? 2. The Inter Party Advisory Committee was established in 2003. Are you aware of the committee? 3. If so, in what ways has the establishment of the committee affected the quality of elections in 2008 and other past elections? 4. In what ways can the election administration in Zimbabwe be improved? 5. What political factors may prevent the successful passage of these improvements? 6. What political factors may contribute to the successful passage of these improvements? 7. If major reforms are not enacted before the upcoming elections, what effect will this have on the quality of the upcoming elections? SECTION 6: Perception of Election Quality 1. On the whole, how would you rate the freeness and fairness of the last national election, held in 2008? 2. Thinking only about the recent national elections, how much confidence, if any, do you have that the 2008 election results accurately reflected the way people voted in the national elections? 3. Let’s break down the various parts of the 2008 elections. How free and fair were these elections in terms of: 4. The availability of information about candidates and parties? b. The use of violence by candidates or parties? 5. The buying of votes by candidates or parties? 6. The counting and reporting of results by the ZEC? 7. For each of the following, please tell me how effective or ineffective you think these 47 institutions were in ensuring that the national elections were free and fair. a. International observers b. Independent Zimbabwean observers c. Zimbabwe Electoral Commission (EC) d. Compared with other elections in Zimbabwe that you are aware of, would you say that the quality of the April 2008 national elections was better, the same, or worse than the others (1999 and 2004)? 8. In your estimation, how free and fair will the upcoming national elections be? 9. In your view, what should be done to ensure that the next elections are free and fair? 48 CHAPTER 3 THE CAUSES OF ELECTORAL ADMINISTRATIVE REFORM IN AFRICA 3.1 Introduction In 2004, President Mugabe of Zimbabwe enacted legislation establishing the Zimbabwe 1 2 Electoral Commission (ZEC) with considerably more autonomy and organizational capacity 3 than the previous electoral administrative institutions. If we assume that President Mugabe was intent on preserving his power and ability to win future elections, what factors influenced the establishment of the ZEC? The reform of electoral management bodies has not been limited to Zimbabwe. Important changes in the rules shaping electoral administration have been carried out in many African 4 5 6 countries including Ghana (1993 and 1996), Senegal (2005), South Africa (1993), and also more recently in Kenya (2010) and Nigeria (2011). Indeed changes to electoral administration 1 For instance, the reforms reduced the power of the executive to appoint EMB commissioners. The president was now mandated to appoint the chairperson in consultation with the Judicial Services Commission, and the four other members were appointed by the president from a list of nominees submitted to the Parliamentary Committee on Standing Rules and Orders (Zimbabwe Electoral Commission Act 2004; See also Wall 2006: 33-36). 2 The Zimbabwe Electoral Commission is responsible for organizing and supervising all national elections, supervising voter registration and compiling the electoral register, developing and distributing voting materials, conducting voter education, and observer accreditation (Zimbabwe Electoral Commission Act 2004; See also Wall 2006: 33-36). 3 Before 2004, three bodies were responsible for election management: The Delimitation Commission, The Electoral Supervisory Commission, and the Registrar-General Elections (Wall et al, 2006). 4 For a detailed analysis of the reform of the Electoral Commission of Ghana see GyimahBoadi (1999). 5 For a detailed analysis of the reforms in Senegal see Lopez-Pintor (2000, 186-194) and Wall et al (2006, 110-113). 6 For more information on South Africa see Anglin (1995). 49 have occurred frequently in many other countries including notable cases in the developing 7 8 9 world such as Mexico, Jamaica, India, and Bosnia 11 democracies, such as Australia 10 and the United Kingdom. as well as changes in more mature 12 These examples of reforms in EMBs present us with an empirical puzzle: Why would the incumbent government establish an independent EMB, or increase the institutional autonomy or capacity of an existing EMB when doing so may reduce the incumbent’s control over the administration of elections and potentially undermine his re-election prospects? This chapter explores the factors that make incumbents more inclined to enhance the autonomy and capacity of the EMB in the context of regimes in Africa. As pointed out in Chapter 1, EMBs are of utmost importance to political elites because these institutions have jurisdiction over several key aspects of the electoral process (such as voter registration, polling, counting of ballots, and tabulation of results). 13 As a result, rules that shape the design of EMBs may have direct implications for the extent to which the electoral playing field is level as well as the degree to which procedural irregularities are minimized. Therefore, it is safe to assume that 7 8 9 See Magaloni (2006a). See Sives (2009). See Wall 2006: 192-95. 10 11 12 13 See Wall 2006: 196-9 See Kelly (2007). See James (2012). For a detailed description of the tasks involved in the pre-election, election day and post election stages of the election see Pastor (1999: 8-9) and Wall et al. (2006). 50 the more control the incumbent has over the EMB, the greater the incumbent’s ability to use the 14 institution to manipulate the election process and ensure victory at the polls. Yet, there have been many instances in which incumbents have taken the decision to formally relinquish control of the EMB by initiating changes that either create a new EMB or considerably improve the existing body’s autonomy and capacity. In doing so, incumbents were able to signal their commitment to making the administration of elections impartial and insulated from executive influence. In some cases, such as Ghana (1996) and South Africa (1999), changes in the formal rules resulted in the improved performance of the electoral management body during elections. In other cases, such as Zimbabwe (2005) and Cameroon (2000), the incumbent agreed to reforms but blocked the successful implementation of these institutional innovations or used other strategies of electoral fraud and manipulation to hold onto power (Levitsky and Way 2010; Trebilcock and Chitalkar 2009). Finally, there are other cases, such as Uganda (2006), where the incumbent has been able to withstand pressures for reform and maintain the status quo. In this chapter, I seek to answer three main questions: 1. What are the determinants of electoral administrative reform in Africa? 2. What factors make some governments more likely to implement reforms that enhance the autonomy and capacity of EMBs and other governments more likely to maintain the status quo? 3. Are the factors that influence the likelihood of capacity-enhancing reforms similar or different from those that influence autonomy-enhancing reforms? 14 Indeed, there are other methods of electoral manipulation (such as voter intimidation, violence against opposition candidates, and vote-buying) that may also be used along with electoral fraud (Levitsky and Way 2007; Schedler 2002). However, I suggest that incumbent control of the EMB is paramount. 51 First, I define electoral administrative reform as the change in the formal rules that influence the design of EMBs. I distinguish between two dimensions of EMB design—EMB autonomy and EMB capacity. Next, I propose a theoretical framework, grounded in historical institutionalism, which associates election administrative reform with the strategic and dynamic interaction among domestic elites (incumbent and opposition) and international actors along with the path dependency of existing institutional rules. Specifically, I suggest that incumbents are more likely to implement significant reforms that enhance EMB autonomy and capacity when the costs of maintaining the status quo outweigh the benefits. I highlight three factors that influence incumbents’ cost-benefit calculations: 1. Strength of domestic opposition relative to the incumbent and the ability of the opposition to credibly pressure the incumbent; 2. Regimes’ vulnerability to external pressure and the commitment of international actors to credibly pressure the incumbent; 3. The extent to which EMB reforms will influence the incumbents’ electoral prospects. To put it more simply, I argue that incumbents are more likely to increase the autonomy and capacity of EMBs when they face credible and enduring pressures from both opposition parties and the international donor community. Within the domestic arena, reform is more likely when a coalition of oppositional political elites supported by key elements of civil society emerges and credibly pressures the incumbent to enact reform. Opposition strategies include public protests and election boycotts. From the international perspective, the likelihood of EMB re- form is enhanced when the international donor community threatens to withhold foreign aid and impose political and economic sanctions in the absence of reform. The international donor communities can influence the struggle for reform by providing resources to empower and 52 legitimize the domestic opposition. Finally, incumbents consider the extent to which EMB reforms can influence their ability to gain office. Incumbents are more likely to pursue significant re- forms when they are certain of winning future elections, or when there are other more effective measures of manipulating elections. The strategic interaction of domestic and international actors only tells one side of the story. While the possibility of EMB reform is always there, there are some periods when reform is more likely than others. Scholars refer to these windows of opportunity as “critical junctures,” “tipping points,” or “exogenous shocks”. What is important is that during these periods, the regime is more vulnerable and the resulting instability increases the opportunity for institutional change. While there are numerous precipitating factors that make the possibility of reform more likely, I focus on loss of regime legitimacy that follows extremely fraudulent elections. These elections often involve extensive incumbent manipulation of electoral administration, gross administrative irregularities, as well as more general problems such as electoral violence and intimidation. To put it more succinctly, bad elections lay bare the need for immediate reform. Finally, this theoretical framework highlights the possibility of periods of stability and incremental changes in electoral administration. I suggest that when political elites consider the EMB a legitimate institution significant institutional change is less likely and path dependency sets in. As a result, significant rule change becomes increasingly difficult both because there are fewer demands for changes from opposition elite and incumbents are less likely to accede to reforms. Nevertheless, there remain opportunities for more gradual change during periods of continuity through changes in the interpretation, implementation and enforcement of existing rules. 53 I examine these theoretical propositions by conducting a longitudinal comparative analysis of four African countries: Ghana (1992-2012: 6 election periods), Nigeria (1999-2011: 4 election periods), South Africa (1994-2009: 4 election periods), and Zimbabwe (2000-2008: 4 election periods). 15 I conduct an in-depth analysis of each election-period by identifying instances of EMB reform (policy stability) and examining whether various factors relating to domestic and international pressures and more general indicators of the political context may have influenced the likelihood of changes in election administration. To do this, I rely on multiple sources of evidence including content analysis of national laws, scholarly analysis of the election period, and policy-based reports by international observers. I also include perspectives from 88 interviews conducted in the four countries among election administrators, leaders of ruling and opposition parties and civil society groups, journalists, academicians, and leaders of international organizations present in each country. The findings from the analyses provide the basis for the following conclusions: 1. The scope and frequency of reforms vary across election periods over time. However, changes in capacity-related reforms occur more frequently than changes in autonomy. 2. The domestic political balance of power between political opposition and the incumbent is crucial to explaining the likelihood of policy innovations in EMB autonomy and capacity. 3. The international community played a decisive role in the reform process by imposing sanctions (Zimbabwe), engaging in diplomatic arm twisting (Nigeria), and providing financial and technical assistance that helped bolster the capacity of oppositional parties and domestic civil society organizations. 15 I detail the case selection rationale in the Methods Section (3.3). For a more comprehensive explanation see Chapter 2. 54 4. Fraudulent elections facilitated critical junctures that undermined regime stability and enhanced the likelihood of EMB policy change. 5. In Ghana and South Africa EMBs have become institutionalized and have not experienced significant changes to institutional autonomy. I organize the remainder of the chapter as follows. First, I begin by reviewing the relevant literature on historical institutionalism and electoral governance. I then present the theory of EMB reform, which is followed by a brief outline of the methodology and present the main findings of the comparative analysis. Finally, I conclude with a discussion of the findings and their main implications. 3.2 Literature Review and Theory 3.2.1 Historical Institutionalism: Explaining Institutional Change New institutionalism is one of the dominant frameworks for explaining the choice and development of political institutions (Hall and Taylor 1996; Pierson 2000a; Steinmo, Thelen and Longstreth 1992; Thelen 1999). 16 There are three prevailing variations of the “new” institutionalism approach: rational choice institutionalism, historical institutionalism, and sociological institutionalism (Hall and Taylor 1996). Of the three approaches, historical institutionalism is most suited for the assessment of electoral administrative reform and is the main theoretical frame that I utilize in the chapter. 16 I employ North’s (1990) inclusive definition of institution as formal and informal “rules of the game” that provide both constraints and incentives for human interaction and behavior. See also Peter Hall’s definition of institutions as “the formal rules, compliance procedures, and standard operating practices that structure the relationships between individuals and various unites of the polity and economy”(1986: 16). 55 Historical institutionalism appropriately combines both features of sociological 17 and 18 rational choice institutionalism : that is, both culture and rational approaches are employed to explain the interplay between institutions and individual action. At the core of the historical institutional analysis are the concepts of power asymmetries, path dependence, critical junctures, and compliance (Mahoney and Thelen 2009). Scholars have incorporated all four concepts to explain periods of institutional stability as well as periods of significant and incremental change. According to Thelen and Mahoney (2009: 8), institutions are “distributional instruments laden with power implications.” Institutions reproduce the power asymmetries within the society, thereby privileging certain coalitions of interest groups over others. In other words, institutions create winners and losers. The privileged groups (winners) have an incentive to preserve the viability of the institution over time and maintain the status quo (Thelen 1999). Conversely, losers are incentivized to initiate reforms to the institution to ensure that the institutional benefits are more equitably distributed. Path dependence refers to the process through which institutional development is guided by the legacies of past institutional configurations and structural factors (Pierson 2000b; Thelen 1999). 19 Institutions remain relatively stable over time (that is they follow a specific path) because of the ways in which they induce various “feedback mechanisms” that reinforce the 17 In the sociological variant of institutionalism, existing institutions represent a schema of culturally relevant norms and practices that informs actors’ decisions to initiate institutional change. 18 Rational choice institutionalists take a micro-level approach to the explaining institutional choice and development. For them, institutions are developed and maintained because of the efficiency-enhancing benefits that they provide (Hall and Taylor 1996; Shepsle 1989). 19 Thelen (1999) also highlights coordination effects as a second mechanism in which institutions, when established, cause actors to formulate strategies that reflect but also reinforce the institution. 56 functionality and legitimacy of the institution for the coalition of actors that benefit from the status quo institutional configuration. Despite the stability induced by these mechanisms, institutional change is still possible either because of institutional inefficiency or as a result of particular processes that undermine the foundations of the feedback mechanisms. These opportunities for significant change emerges during critical junctures, what Thelen defines as “crucial founding moments of institutional formation that send countries along broadly different institutional paths” (Thelen 1999: 387; see also Capoccia and Kelemen 2007). Critical junctures arise through either endogenous or exogenous shocks to the system. During these windows of opportunity, there is significant modification of macro-structural factors, political processes, and the distribution of power among actors. Institutional rules are now more susceptible to transformation. Prevailing structures and institutional legacy give way to the strategic interaction and agency of the institutional winners and losers. While periods of significant institutional change are produced through shifts in the balance of power of political actors during critical junctures (exogenous and endogenous shocks), change can also occur more gradually. Gradual or incremental change is facilitated through the process of compliance. Compliance is a relatively new addition to historical institutional- ism, and is predicated on the belief that the interpretation, implementation and enforcement of institutional rules also have distributional consequences (Thelen and Mahoney 2009). As a result, there will also be struggles among various actors about the interpretation, implementation, and enforcement of institutional rules. The outcome of these battles has implications for the institutions’ ability to change gradually over time and for the institutions’ overall performance (ibid). 57 3.2.2 Explaining EMB Reform: A Historical Institutionalism Approach Scholars of electoral administration have applied these insights from historical institutionalism to understand the design of EMBs (Mozaffar 2002; Gazibo 2006; Lehoucq and Molina 2002). In particular, Shaheen Mozaffar’s (2002) path-breaking research on the determinants of electoral governance patterns in sub-Saharan Africa represents the first systematic study of EMB design in emerging democracies. Mozaffar argues that institutional legacies of colonial governance and post-colonial neopatrimonial regimes, ethnopolitical cleavages, and the type of negotiations during regime transitions are crucial determinants of the formal-legal autonomy of EMBs in Saharan Africa. 20 Mozaffar finds that countries with low levels of political competition during an authoritarian regime were less likely to create autonomous EMBs, while countries that had any negotiations over the design of the EMB were more likely to create autonomous EMBs. In this way, Mozaffar highlights the interaction between path-dependent trajectories of institutional legacies with the voluntarisms and agency of leading political actors. However, an important limitation of Mozaffar’s research is that it only assesses the factors influencing the initial institutional design of EMBs during the transition period; there is little systematic assessment of how the main explanatory factors influence the evolution of EMB autonomy over time. Moreover, the research employs explanatory indicators that privilege longterm structural factors instead of more proximate indicators of the strategic interaction of elite 20 Mozaffar distinguishes between three types of electoral management body designs: autonomous, semi-autonomous, and non-autonomous with respect to the institutional location of the electoral management body relative to executive. 58 actors, thereby diminishing the true effect of the strategic interaction of transition actors (Gazibo 2006). Another important advancement in the study of EMB reform came with Mamoudou Gazibo’s research on EMB autonomy (2006). Gazibo builds upon Mozaffar’s research by arguing that variations in EMB autonomy and “efficacy” are a function of the distribution of power among elites during democratic transition periods. As he suggests, “[W]hen a commission rises out of circumstances in which no political wing is powerful enough to control the whole process, the chances are greater that this com- mission will enjoy a more substantial degree of autonomy” (2006: 628). Utilizing a historical institutionalism framework, Gazibo explains EMB institutional design as the interaction between the effects of longstanding institutional legacies with the explicit focus on the elite-power relations that guided the democratic transition process. Unlike Mozaffar, Gazibo’s concept of autonomy highlights the empirical realities, rather than focuses on the formal-legal provisions. 21 To predict variations in the empirical autonomy, Gazibo relies on the allocation of power among domestic political elites (incumbents/outsiders) during the 22 democratization process and outlines five models of institutional design: the consensual; 21 the For Gazibo, this conceptual distinction is important because EMBs “may be formally independent but empirically powerless, while others may be supposedly subordinate but over time be able to strengthen their capacity far beyond legal prescriptions” (620). 22 The consensual model emerges when both incumbents and opposition are equally engaged in the institutional formation and results in an autonomous EMB potentially susceptible to political manipulation. 59 incumbent-led interactive; the outsider-led interactive; models. 23 the imposed; and the managed 24 Gazibo’s model of domestic political power asymmetries between oppositional and incumbent actors provides a springboard for my inquiry into the dynamics of EMB creation and reform. Despite Gazibo’s excellent grasp of the different configurations of domestic power struggles his research generally ignores the influence of the international community as well as the strategic battles over EMB capacity. In this chapter, I make the case that institutional reform in the context of Africa cannot be adequately explained without incorporating 1) the effect of international community on incumbents’ strategic decisions, and 2) the effect of international community on enhancing the power of oppositional groups. I also seek to make conceptual modifications to Gazibo and Mozaffar’s research. I conceptualize electoral administrative performance more broadly to include, not only institutional autonomy but also institutional capacity. As I emphasized in Chapter 1, this conceptual refinement is important because autonomy and capacity represent two distinct aspects of EMB performance. The extent to which both are pursued (concurrently or independently) has important consequences on the reform process. Finally, I suggest that changes in the autonomy and capacity of the EMB can occur gradually over time through the ways in which institutional rules are interpreted and implemented. 23 The incumbent-led interactive and outsider-led interactive models describe the forging process in which the incumbent and the opposition are involved, but either the incumbent or opposition has the advantage resulting in a EMB with lower (higher) levels of autonomy, with the potential of institutional autonomy being reversed with alternations in power. 24 The managed and imposed models highlight the impact of the incumbent or opposition unilaterally designing the EMB. The managed process produces EMBs that are appendages of the government, while the imposed induces high levels of autonomy. 60 3.2.3 Explaining EMB Reform: Theoretical Framework As noted in the introduction of this chapter, we are confronted with an empirical puzzle in which some African incumbents introduce reforms to enhance the autonomy and capacity of election management bodies (EMBs) when doing so may potentially undermine the incumbents’ ability to rig elections in their own interests. I contend that incumbents are more likely to implement significant reforms that enhance EMB autonomy and capacity when the costs of maintaining the status quo outweigh the benefits. I highlight three factors that influence incumbents’ cost-benefit calculations: 1) the strength of domestic opposition relative to the incumbent and the ability of the opposition to credibly pressure the incumbent; 2) the regimes’ vulnerability to external pressure and the commitment of international actors to credibly pressure the incumbent; and 3) the extent to which EMB reforms will influence the incumbents’ electoral prospects. 3.2.4 Impact of EMB Reform on an Incumbent’s Electoral Strategy The main dilemma that incumbents encounter when faced with demands for electoral administrative reform is the effect that these reforms can have on their own reelection prospects. As shown in the Table 3.1 reforming the EMB: 25 there are four options open to the incumbent who considers 26 1. Enhance both autonomy and capacity 2. Enhance autonomy with no change in capacity 25 26 All figures and tables are placed in the Appendix at the end of each chapter. The incumbent could also implement reforms that reduce the capacity or autonomy of the EMB, yet I do not consider this formally in the model for parsimony. However, I acknowledge that this is a possibility. For example during the 2002 presidential elections in Zimbabwe the incumbent passed legislation to diminish the capacity and autonomy of the EMB. 61 3. Enhance capacity with no change in autonomy 4. Maintain the status quo with no change in autonomy or capacity The first option—increasing both capacity and autonomy– provides the most credible signal to both international and domestic actors. 27 However, when compared to the other possibilities, it most effectively minimizes the incumbents’ ability to manipulate the EMB and elections more generally. The second option—enhancing autonomy relative to capacity—may also pro- vide a credible signal to domestic and international actors; however, the improvements garnered through a more autonomous EMB may be undermined if the incumbent can take advantage of the limitations in capacity. The last option for reform involves increasing the capacity of the body without improving autonomy. In my estimation this may appear to signal a commitment to improving the EMB operations, but in actuality, it potentially gives the incumbent greater control over the election process as he still possesses the ability to influence the EMB, and with its increased capacity the degree of manipulation can be more widespread than previous elections. Additionally, incumbents also consider the scope of the reforms. As a result, incumbents should be more inclined to pursue reforms with limited scope (such as those that modify existing institutional rules) than those that have broad implications for electoral management (such as the creation of a new electoral management institution). As highlighted previously, incumbents are more likely submit to EMB reforms when they face demands for reforms from a strong opposition and international actors with significant 27 I acknowledge that the reference to both capacity and autonomy gives no indication of the types of reforms that are enacted by the incumbent. For instance, some reforms to EMB autonomy may have more consequences for the incumbents’ ability to manipulate elections than others. 62 economic and political leverage, and international actors with strong links with domestic agents of change. 3.2.5 Strength of Opposition Oppositional actors can make it more difficult for the incumbent to resist institutional reform through three main strategies: 1) staging electoral boycotts (or threatening to boycott), 28 mobilizing popular protests, and 3) developing legislative coalitions to enact electoral re- form. To begin, oppositional boycotts can induce reform by undermining the legitimacy of the current regime (because elections are no longer seen as competitive); they can also serve as a focal point for mobilizing domestic and international support for the reform initiative (Bratton 1998; Lindberg 2006a; Beaulieu and Hyde 2009). However, in Africa the effectiveness of boycotts (or credibly threatening boycotts) is limited due to the difficulty of creating a coordinated and cohesive opposition 29 and the inability of parties that have boycotted to spearhead reform through the legislative process (Rakner and van de Walle 2009). A second recourse for opposition groups is to coordinate with civil society groups 30 to mobilize mass protests movements against the government (Kuntz and Thompson 2009; Norris 2011). Protest movements have been especially effective during the transition processes in sub28 Albert Hirschman’s (1970) research on exit, voice, and loyalty provides a theoretical frame for understanding the main strategies of opposition elites during the reform process. The mobilization process reflects the voice-based response to the unequal electoral administrative context. Meanwhile, elites can exit from the elections by deciding to boycott. But in many regimes due to the ability of leaders to punish and or co-opt disgruntled elites, many opposition groups maintain their loyalty to the regime 29 van de Walle argues that opposition cohesion is a function of the likelihood of opposition victory, level of democracy, and certain structural factors such as institutional context (tworound majority voting), history and culture, ethnic pluralism, socio-economic development, and the degree of international pressure and support for political parties (2006: 82-92). 30 Bratton and van de Walle believe that “civil society groups must be adequately organized into primary associations” (1992). 63 Saharan Africa in countries such as Benin, Nigeria, South Africa, Cote d’Ivoire (Bratton and van de Walle 1997) and Eastern Europe (McFaul 2002; Bunce and Wolchik 2009). More recently, opposition elites have organized massive mass protests in Nigeria (2010) to promote electoral reform (Aiyede 2012). 31 A third potential strategy for oppositional parties is to form coalitions within the legislature to initiate reform (Barkan 2009). This is, however, constrained in many African polities because of the (1) dominant role played by presidents in initiating legislation; 32 (2) the limited strength and cohesiveness of opposition parties in legislatures — due mainly to the high fractionalization of African party systems; 34 oppositional candidates; 33 (3) the president’s use of clientelism to co-opt and (4) the low capacity of African legislatures to initiate reforms. 35 Based on these propositions I contend that, • (H1): The incumbent is more likely to institute enhancing EMB reforms when opposition parties are strong and pressure the incumbent to enact reforms • (H1a): The incumbent is more likely to institute enhancing EMB reforms when opposition parties boycott elections. • (H1c): The incumbent is more likely to institute enhancing EMB reforms when opposition parties threaten to boycott elections. 31 However, as I will discuss, the opposition’s ability to mobilize mass protests depends on the coercive capacity of the government and its willingness to use force to suppress on opposition movements. 32 See Barkan 2009; Lehoucq and Molina 2002: 25; and Nijzink, Mozaffar and Azevedo 2006. 33 See Kuenzi and Lambright 2005. 34 35 See van de Walle 2003. See Barkan 2009. 64 • (H1d): The incumbent is more likely to institute enhancing EMB reforms when opposition parties mobilize mass protests. • (H1e): The incumbent is more likely to institute enhancing EMB reforms when opposition parties develop strong legislative coalitions. 3.2.6 Incumbent Organizational and Coercive Power Another way of viewing the strength of the opposition vis-a`-vis the executive is by assessing the incumbent’s organizational and coercive power. According to Levitsky and Way, there are two ways to assess the incumbents’ ability to counteract the oppositions’ reform agenda (2010). 36 The first is the coercive capacity of the state or the extent to which the incumbent can use state security agencies to deter mass opposition protests. The second is the organizational capacity of he incumbent that is often gauged through the existence of an incumbent affiliated political party and the cohesiveness of that political party (Brownlee 2007; Gandhi 2008). Based on this I suggest the following, • (H2): Incumbent is less likely to institute enhancing EMB reforms when the incumbent possesses high organizational and coercive power. 3.2.7 Role of International Community The international community represents another important dimension of the power asymmetries associated with election management bodies. International actors, especially those that place a priority on promoting democracy (such as Western donors, NGOs, and international observer organizations) have been endorsing greater autonomy and capacity of electoral 36 Other scholars have highlighted the impact of incumbents’ ability to maintain elite and mass based patronage 65 management as an essential prerequisite of credible elections (Carothers 1999). Consequently, international actors may increase the costs to the incumbent of maintaining the status quo through a variety of strategies. First, Western governments and multi-lateral organizations may promote democratic reforms and use “carrots”—such as increased foreign aid—or “sticks”— including diplomatic and economic sanctions—to encourage their implementation (Levitsky and Way 2010; Birch 2011). Second, international actors can provide direct democracy assistance to agents of change, including opposition parties and civil society to enhance their capacity to organize and mobilize popular support (Wright 2009; Hearn 2000; Bunce and Wolchik 2009). Direct assistance may also be extended to EMBs through financial, technological, and infrastructural capacity building, which reduces the direct costs to the incumbent for making capacity-related reforms (Wall, for Democracy and Assistance. 2006). Third, international actors such as international observer groups and international NGO’s can use the bully pulpit during elections to encourage EMB reforms (Hyde 2011; Kelley 2012). Taken together, these strategies reflect insights from the literature on international linkage and leverage. According to Levitsky and Way a country’s political, economic and social integration into the international community has crucial consequences on the strategic actions of 37 incumbent governments and opposition actors (2007; 2010). 37 I am not, however, advancing a Western leverage refers to “governments’ vulnerability to external pressure” that depends on states’ “bargaining power” relative to western powers or their ability to withstand political and economic sanctions (2006:48). Western leverage, if effectively applied, raises the costs of maintaining the status quo of electoral institutional structures for incumbent governments. The second dimension of international influence is linkage, which denotes “the density of a country’s economic, political, organizational, social, and communication ties to the West.” Particularly important for this study are the ways in which greater linkage between western governments, NGOs, domestic opposition groups, and civil society organizations can help to reorient domestic preferences associated with international democratic norms, shift the balance of power in favor of oppositional groups through greater resources, and enhance popular legitimacy of domestic groups (Way and Levitsky 2006). 66 theory that gives primacy to international factors, or overestimates their influence, because there are many features of African politics that limit the power of linkage and leverage in effecting 38 reform (Bayart 1993; van de Walle 2003; Bratton and van de Walle 1997: 31). • (H3a): The incumbent is more likely to institute enhancing EMB reforms when international actors place credible conditionalities on the implementation of EMB reform. • (H3b): The incumbent is more likely to institute enhancing EMB reforms when international actors provide resources to empower and legitimize the domestic promoters of reform (opposition and civil society) • (H3c): The incumbent is more likely to institute enhancing EMB reforms when international actors provide resources to improve the capacity of EMBs. 3.2.8 Critical Junctures and Path Dependency Critical junctures represent windows of opportunity in which agents of change (opposition actors, domestic and international democracy promoters) are more likely to enact significant re- forms to electoral administration. While there are numerous exogenous and endogenous factors that can affect the distribution of political power and the feedback mechanisms guaranteeing institutional ability, I focus on just one. I contend that changes in electoral administrative institutions are more likely to occur following fraudulent elections, which vitiate the domestic and international legitimacy of the regime (Kuntz and Thompson 2009). • H4: The likelihood of EMB reform increases following fraudulent elections. 38 Many incumbents have been able to withstand economic and political sanctions and find alternative means of preserving power and autonomy (Bayart 1993). While other governments have symbolically adopted reforms without any substantive changes (van de Walle 2003). 67 3.3 Data and Methods To identify the determinants of EMB reform, as well as the relevant causal mechanisms, I propose a small-n longitudinal comparative analysis of four African countries Ghana (19902012), Nigeria (1999-2011), South Africa (1994-2009) and Zimbabwe (2000-2012). The unit of analysis is the election period, which starts from six months after the last election (time t-1) to six months after current election (time t). 39 There are a total of 18 election-period cases across the four countries. As highlighted in Chapter 2 of the dissertation, the selection of each country is guided by the need to have significant variation in EMB capacity and autonomy. Cases were also selected based on similarities such as colonial heritage, post-colonial regime type, size of population, ethnic diversity, and level of economic development. 40 Table 3.2 displays the variation in the balance of EMB autonomy and capacity in each election period across the four countries. For instance, during the period under investigation Ghana’s electoral management body was established with relatively low levels of autonomy and capacity (1992 election period) but both autonomy and capacity increased by the 2000 election. On the other hand, Nigeria maintained very low levels of EMB autonomy and capacity (19992007); 41 but demonstrated significant improvements in 2012. Zimbabwe began in 2000 with low levels of autonomy but had moderate/high levels of administrative capacity (2000-2008). Finally, 39 To put it more simply, the election period for an election in Ghana in 2000 starts six months after the last election in 1996 and ends six months after the 2000 elections. 40 41 Please see Chapter 2 for a complete description of the case-selection criteria. For instance, according to the PNDC Law 271, the executive selects members of the interim National Election Commission without any consultation with opposition parties. 68 South Africa began with very high levels of autonomy and relatively low capacity that increased over time. 3.3.1 Qualitative Methodology I employ causal process observation (Brady and Collier 2004; George and Bennet 2009) as the primary method to examine the proposed theory of EMB reform. According to George and Bennet, two leading proponents of mixed methods, the “process-tracing method attempts to identify the intervening causal process—the causal chain and causal mechanism—between an independent variable (or variables) and the outcome of the dependent variable” (2009:206). Causal process observation has become one of the most utilized analytical methods of studies on institutional change (see for instance, North 1990; Skocpol 1979; Mahoney and Thelen 2009). More specifically, I conduct analyses of each election period and utilize a variety of sources to corroborate or disprove the theory of EMB reform. 42 First, I identify instances of formal or informal changes in the level of EMB autonomy and capacity. Second, I examine the domestic events preceding reform including evidence of opposition strength, incumbent organizational power, international linkage and leverage and evidence of critical junctures. 3.3.2 Measuring EMB Reform The main object of explanation is the extent of EMB reform during the election period. I define electoral administrative reform as the change in the formal rules that influence the design of EMBs. I distinguish between two dimensions of EMB design—EMB autonomy and EMB capacity. The autonomy and capacity of the EMB can be altered formally (or informally)— 42 It is important to note that the purpose of the analysis is not to recount all the details of the election periods, but instead to highlight the possible factors that might have increased or decreased the likelihood of reforms. 69 through constitutional, statutory, or regulatory changes. Furthermore, I classify the scope and relevance of reforms as major, moderate, and/or minor on the basis of whether reforms are defined in the constitution, legal statutes, or EMB regulations respectively. There are numerous indicators of EMB autonomy and capacity. For the purposes of this chapter, I focus on four main indicators – institutional, professional, partisan, and financial autonomy and five main indicators of capacity – professional, administrative, election process, stakeholder cooperationconflict resolution, legislative, and sanctioning; see Table 3.3 for main autonomy and capacity indicators. For instance, EMB autonomy can be improved by changing the institutional location of the EMB (that is from being located in a government ministry to being an independent body located outside the direct purview of the executive). 43 Other formal indicators of EMB autonomy include the nature and inclusiveness of the appointment process for EMB commissioners or the professional requirements for commissioners. Similarly, we may observe changes in capacity through improvements in voter registration systems, rules on how citizens should vote, and improvements in the system used to count ballots and announce results. 3.3.3 Domestic and International Balance of Power There are various ways to gauge the balance of power between the domestic and international supporters and challengers of EMB reforms. From the perspective of opposition parties, there are three indicators that may highlight their commitment to pursuing EMB reform: 1) op- position boycott (or threat of opposition boycott); 2) opposition-organized mass protests; 43 This change was carried out in Zimbabwe in 2005 when the Zimbabwean Electoral Commission, a constitutionally independent body, replaced three separate bodies that were controlled by the government (Delimitation Commission, Election Directorate, and Electoral supervisory Commission). 70 and opposition seat share in the legislature. I measure incumbent organizational and coercive capacity through the legislative seat share of the incumbent and the scope and extent of executive control over state security apparatus. From the perspective of international actors, I assess their commitment and ability to promote EMB reform through evidence about international sanctions applied to the country and electoral support for either EMBs or domestic opposition. 3.3.4 Critical Junctures To effectively capture the critical junctures, I focus on the extent to which the EMB was used in the previous election to perpetuate fraud, as well as the overall quality of previous elections. Finally, I consider whether the EMB reforms were implemented during an overhaul of the constitution or reforms followed a regime transition. 3.4 Empirical Findings 3.4.1 Ghana (1992-2012): A Model for Africa When compared with the other countries in the sample, Ghana’s electoral administration has experienced the most significant institutionalization of autonomy and capacity following the reintroduction of multi-party elections in 1992. 44 There were two main trends in EMB policy reforms: 1) the 1992-1996 election periods, and 2) the 2000-2012 election periods. The first two election periods —1992 and 1996— provide significant support for the proposed theory. During these two election periods, Ghana’s former military ruler, Flt. Lt. Jerry Rawlings, was instrumental in improving the autonomy and capacity of electoral administration. 44 45 45 Some of See Table 3.5 for a list of the EMB reforms in Ghana 1992-2012. Flt. Lt. Jerry Rawlings had assumed power in a coup in 1981 and developed the Provisional National Defense Council (PNDC). According to Bratton and van de Walle (1997) 71 these changes included the establishment of a new, constitutionally independent electoral management body, comprised of professional EMB commissioners who enjoyed lifetime tenure and insulation from political control (Jeffries 1998; Gyimah-Boadi 1999). numerous innovations in the voter registration process, results-announcement procedures 49 47 46 48 voting operations There were also and counting, and (Gyimah-Boadi 1999: 108-109). Finally, Ghana’s electoral management body also incorporated an informal mechanism for promoting inter-party communication and dialogue over electoral administrative matters known as the Inter-party Advisory Committee (IPAC) (Ayee 1997: 426). 50 Why did the incumbent President Jerry Rawlings agree to these sweeping reforms? According to the theory on EMB reform, incumbents who face very high costs for maintaining the status quo from both the domestic actors and international community are more likely to implement democratic changes to election administration. There are three factors that influenced President Rawlings’ commitment to implement enhancing EMB reforms (1992-1996). (See Table 3.4 for a summary of variation in EMB autonomy reforms and the main causal factors for election periods across the four countries under investigation). “both donor pressure and popular discontent compelled the regime” to carry out a managed transition towards democratic rule. 46 It is important to note that the PNDC developed the Interim National Electoral Commission (INEC) to supervise the 1992 referendum on the draft constitution, the founding executive, and parliamentary elections (Jeffries and Thomas 1993). Rawlings appointed all the members of INEC, without any objections by the opposition (Jeffries and Thomas 1993: 338). 47 Including the introduction of voter ID cards (1996). 48 49 Including the introduction of transparent ballot boxes (1996). Including, counting ballots at polling stations and posting results at polling stations (1996). 50 Commenting on the significance of IPAC during the 1996 elections, Gyimah-Boadi, a leading political scientist notes “IPAC played a crucial role in building considerable consensus on the electoral process and reforms towards the 1996 elections” (1999: 112). 72 First, following the founding 1992 Presidential elections, the main opposition party National Patriotic Party (NPP) complained bitterly about the extensive electoral fraud and manipulation of the EMB (the Interim National Electoral Commission) (Boahen 1995a; Oquaye 1995). In response to the allegations of fraud, the NPP mobilized their supporters to conduct nation-wide protests and boycotted the upcoming parliamentary elections scheduled for the following month. 51 The allegations of fraud following the 1992 presidential elections, the successful opposition boycotts, and the mass-protests provided a critical juncture for reform. Although Rawlings was convinced that he was the rightful winner of the election 1992, he was still concerned that the allegations would damage his popular image and reputation as a democratic reformer (Jeffries 1998). The 1996 elections provided an opportunity for him to repair his image and prove that he could win the presidential elections with a completely reformed electoral system. 52 A second possible reason for the incumbents’ embrace of reform was that the opposition boycotts in 1992 tarnished the international image of the Rawlings administration. Although the 51 The main opposition party, New Patriotic Party, provided three main reasons for the boycott (Boahen 1995). First, elections were rigged in favor of the incumbent, preventing the true democratic preferences of the electorate to be realized. “The second and even more important reason is our expectation that if we boycotted the parliamentary elections in protest against the rigging of the presidential election, the international community would prevail upon Rawlings in the interest of equity and fair play to cancel the latter and start the whole democratic process again” (Boahen 1995: 279). Lastly, was the belief that participating in the parliamentary elections would further legitimize the Rawlings’ regime. 52 Personal interview with Ghanaian Academic Accra, Ghana, May 2010. 73 53 international community did not formally sanction the government for the fraudulent elections, many commentators believed that Rawlings support of a more politically insulated and administratively effective EMB was driven by his concern for his international reputation and the fear that any further allegations of electoral corruption might derail Ghana’s economic reform program (Jeffries 1998; Levitsky and Way 2010). 54 Third, the financial and technical costs of instituting reforms were significantly reduced through assistance from the international community. For instance USAID contributed $8.5 million through the Supporting the Electoral Process (STEP) project, which was instrumental in building the EC’s capacity to register 9.2 million Ghanaians (Lyons 1997:7). Furthermore, other international partners such as Denmark and the UK funded key voting materials, trans- parent ballot boxes, and private voting screens. 55 In sum, President Rawlings displayed a genuine commitment to promoting the establishment of an autonomous, well resourced, and powerful electoral commission during the 1996 election period. He did this by collaborating with opposition parties, seeking assistance from the international donor community, and providing the EC with the necessary resources needed to build its capacity, while at the same time preserving its political autonomy (GyimahBoadi 1999: 114; Levitsky and Way 303; Jinadu 2011). 56 53 From the perspective of Levitsky and Way (2010), Rawlings faced very little international pressure following the 1992 elections. In fact western donors pledged $2.1 billion in aid. 54 According to Jefferies (1998), during the 1990s Ghana faced economic crisis and needed international support from donor agencies to prevent further economic disaster. It was therefore important for the government to toe the line with the international community. 55 Denmark spent $3 million, while the UK spent $0.8 million (Lyons 1997). 56 The Chairman of the EC in Ghana, Dr. Afari-Gyan, admitted that there were no specific instances that President Rawlings attempted to undermine the independence of the commission. 74 Since 2000, Ghana’s electoral administration has not experienced any significant autonomy related reforms and only moderate reforms in capacity that have been proposed and implemented by the electoral commission (See Table 3.4). In many ways during this period the EC begins to develop a path dependency and establish institutional stability. There are three explanations for this period of relative institutional stability and gradual improvements in capacity. First, Ghana’s Electoral Commission has established a reputation for political independence and professionalism by presiding over five consecutive elections, all of which were considered free and fair and two that resulted in victory for the opposition (Debrah 2011). Consequently, the commission has developed legitimacy and enjoys widespread confidence from members of major political parties, civil society organizations, and the international donor community (Gyimah-Boadi 2009). 57 Second, and more general, is the prevailing democratic political context in Ghana. Since 2000, Ghana’s democracy has become increasingly consolidated: there are a strong rule of law, independent media, an independent judiciary and competitive multi-party politics. This is important as it reduced the possibility for incumbent manipulation of the EMB and electoral irregularities. Finally, the institutionalization and relative freedom from political manipulation, has provided EC executives with discretion to develop appropriate legislation and expertise needed to implement capacity-related reforms, such biometric voter registration in 2012 (CODEO 2012). 3.4.2 South Africa (1994-2009): Forerunner in Electoral Administrative Innovation However, he indicated that the NPP attempted to abridge the EC’s constitutional autonomy on two occasions between 2000 and 2008 (Personal Interview, Accra, Ghana, June 2010). 57 The widespread confidence in the EC was also reflected in personal interviews conducted with political party executives, leaders of NGOs, and representatives of leading international organizations. 75 The development of South Africa’s post-apartheid election administration bears many similarities to Ghana between 1992-2012. 58 During the 1994 and 1999 election periods, South Africans witnessed significant reforms, including the creation of a constitutionally enshrined independent EMB—Independent Electoral Commission (IEC)—with one of the most inclusive appointment processes of EMB commissioners in Africa. 59 South Africans also benefitted from the adoption of cutting-edge technologies in voter registration, voting operations, and counting and tabulation. In subsequent election periods (2004 and 2009), there were no major reforms that influenced the autonomy of IEC, but the IEC continued to propose new legislation, and several new procedures to remedy longstanding challenges with voter registration, conflict management, and results announcement. 60 All things considered, the IEC represents one of the best examples of political institutional innovation in Africa and indeed in the developing world (Trebilcock and Chitalkar 2009). The decision to adopt an independent model of electoral administration during the founding elections in 1994 was not inevitable, as during the apartheid regime the Department of Home Affairs (DHA) was responsible for election management (Padmanabhan 2002). 61 One reason for the change was that during the negotiated transition, the power of the incumbent 58 59 See Table 3.7 for a list of the EMB reforms in South Africa between 1994-2009. The appointment process of the six IEC commissioners is more inclusive than both Nigeria and Ghana, in which the President makes appointments based on recommendations from the National Assembly based on a list of nominees prepared by different independent institutions, including the Human Rights Commission, Constitutional Court, the Commission on Gender Equality, and the Public Prosecutor (The Constitution of the Republic of South Africa 1996; Electoral Commission Act, 51 of 1996). 60 According to the Electoral Commission Act, 51 of 1996, the IEC is empowered to develop electoral technology, review electoral legislation, and engage in electoral research. 61 Under the apartheid regime the South African elections were administered by a government model of electoral administration through the Department of Home Affairs (Anglin 1995). 76 apartheid government led by the National Party had waned significantly relative to the main opposition African National Congress (ANC) (Kadima and Booysen 2009). This power imbalance in favor of the ANC gave the party more sway over the decisions concerning electoral management. For example, the decision to create an autonomous body to supervise elections was linked ANC’s mistrust of the DHA to credibly and impartially conduct the election as it was staffed by appointees of the former apartheid government. Moreover, most stakeholders demanded a body that fully represented the diversity of political and societal interests in the new democratic South Africa as the DHA was staffed by mostly white South Africans (Anglin 1995; Fick 1998). A second reason was that the international donor community played a crucial role during the 1994 regime transition. The United States and the European Union provided extensive 62 financial and technical support to the IEC (Piper 2005:86). In fact, international actors were so instrumental to the transition process, and the integrity of election administration, that 2 of the 16 seats on the IEC were reserved for members of the international community (Padmanabhan 2002: 1166). Because the transitional IEC was only mandated to supervise the 1994 elections, during the 1999 election period South Africans were once again faced with a decision as to what form of electoral management design they would use to govern elections (Kadima and Booysen 2009: 398). The final decision to create a new independent body possessing enhanced power and authority was influenced by three important issues (Padmanabhan 2002: 1175). The first relates to the opposition parties’ distrust of the ANC administrations’ ability to impartially govern elections. The 1994 elections created a significant shift in the distribution of political power, as 62 Piper (2005) notes that USAID, a democracy promotion agency that operates within the United States State Department, spent $250 million on the South African elections. 77 the former opposition ANC gained 62 percent of the popular vote (Kadima and Booysen 2009: 428). 63 The second is based on the prevailing lack of confidence of major stakeholders in the DHA’s independence. 64 The third is associated with the pervasive administrative problems that faced the IEC during the transitional elections and the belief that only an independent electoral management body would have the ability to resolve these problems (Padmanabhan 2002: 1175; Piper 2005: 11). 65 Overall, the process of designing the EMB was truly consultative and all major parties submitted reform proposals that emphasized their core concerns (Padmanabhan 2002: 1175). Since the 1999 election period, there have been no major reforms that sought to enhance the autonomy of IEC (Kadima and Booysen 2009). Nevertheless there have been rumblings among major opposition groups for further curtailment of executive involvement in the appointment of commissioners to the Independent Electoral Commission (IEC). 66 Regarding capacity re- forms, South Africans experienced consistent improvements in the 2004 and 2009 63 The National Party and Inkatha Freedom Party received 20 percent and 11 percent of the remaining voter share (Kadima and Booysen 2009: 428). 64 A leading Member of Parliament and executive of the Inkatha Freedom Party shared this perspective (Personal interview, Cape Town, South Africa November 2010). 65 The 1994 election were, generally, peaceful and orderly. However, the IEC was overcome with administrative dilemmas: ranging from the absence of a formal voters register, inadequate training of polling officials, and irregularities in the counting and tabulation process (Elklit and Reynolds 2000: 26; Anglin 1995; Padmanabhan 2002; Piper 2005). Despite, these administrative problems, all major political parties accepted the election results (Elklit and Reynolds 2000). 66 Members of parliament representing some of the main opposition parties in South Africa shared these sentiments, including Congress of the People, Democratic Alliance, Inkatha Freedom Party and the United Democratic Movement. In the view of one South African MP “since multi- party democracy led to the formation of other parties, there has been no restructuring of the process of appointing IEC commissioners that would increase the level of transparency. The IEC and ANC-led government have rejected any attempts to change the membership: 99% of the members are ex-ANC members” (Personal interview, Pretoria South Africa, December 2010). 78 election periods. For instance, in 2004 the IEC increased the number of polling stations, implemented a voter registration verification system (both online and through SMS), and developed a new National Results Center to increase the transparency of the counting and announcement process (IEC 2009). In 2009 it introduced a system to electronically scan results slips from individual polling stations to reduce human error in the recording and transmission of results (ibid). The institutionalization of the IEC bears some similarities with the EC in Ghana. Since the 2004 election period, most election stakeholders including political parties, civil societies, and international donors, have regarded the IEC as a legitimate political institution. 3.4.3 Nigeria (1999-2011): Getting Election Administration Right in 2012, After Three Unsuccessful Attempts. Unlike the other countries under investigation, Nigeria has implemented major and 67 moderate EMB reforms to enhance autonomy and capacity in every election period since 1999. However, between 1999 and 2007 these rule changes did not considerably improve the actual capacity and autonomy of election administration in Nigeria (Jinadu 2011). This was because the EMB reforms were either not effectively implemented or they did not effectively address the susceptibility of Independent National Electoral Commission (INEC) to executive manipulation (Suberu 2007: 97; Alabi 2009: 288). 68 Nevertheless, during the 2011 election period there was a renewed commitment among domestic and international actors to improve the quality of election 67 See Table 3.9 for a list of the EMB reforms in Nigeria 1999-2011. 68 In one of the most comprehensive analyses of election administration in Nigeria since independence in 1960, Adele Jinadu identifies the legacy of politically manipulable EMB as a principal shortcoming of election administration in the country. He explains how during different episodes of democratic rule political actors established nominally independent electoral administrations; the political actors repeatedly utilized these institutions to rig elections. As a consequence electoral administration in Nigeria has a legacy of popular mistrust. 79 administration and the country implemented the most comprehensive package of EMB reforms in its history. Scholars and policy-experts have credited these reforms for the staging of Nigeria’s 69 most credible elections in 2011 (Jinadu 2011; Lewis 2011; Akhaine 2012). I start by examining the period between 1999 and 2007. I highlight three reasons why Nigerian incumbents resisted making significant reforms that could have enhanced the autonomy of the EMB and limited their ability to manipulate election administration. First, unlike Ghana, Nigeria’s opposition parties are relatively weak, regionally fragmented, easily co-opted. 70 Consequently, the opposition was unable to pressure the ruling Peoples Democratic Party (PDP) into making any meaningful reforms (LeVan, Pitso and Adebo 2004). Although opposition parties worked in conjunction with civil society groups to demand reforms before the 2003 and 2007 elections, they did not possess the legislative cohesiveness to effectively shape the legislation passed by the PDP. Even the strategy of symbolic election boycotts in response to widespread election fraud in 2003 did not provide the opposition with enough leverage to significantly change the dynamics of the reform process (Ibrahim and Egwu 2005). Second, because of Nigeria’s economic and strategic importance to the United States and other Western countries, the international community never utilized the threat of economic or political sanctions to leverage improvements in election management. This was the case even as most international donors were aware of the patently fraudulent elections in 1999 and 2003 (Obi 69 In Chapter 7, I engage in an indepth assessment of the consequences of the reforms instituted within the 2011 election period. 70 Interestingly, however, Nigeria has had a legacy of strong civil society groups that coalesced into a vibrant pro-democracy movement during the military rule in the 1980s and 1990s (Edozie 2002). 80 71 2008:81). For Nigeria, the leverage of the international community has been mediated by its strategically important oil reserves. Herskovits explains the link between Nigerian oil resources and U.S. support of the Obasanjo regime: The United States gets some ten percent of its crude oil from Nigeria and hopes to get more in the future . . . believing that [President] Obasanjo is central to these U.S. interests, Washington has supported him, despite human rights abuses and the flawed 2003 elections. (2007: 116) A third reason why many of the reforms were not implemented and why INEC failed to effectively develop procedures that would improve its capacity was because of Nigeria’s legacy of strong and pervasive informal institutions of corruption and clientelism (Joseph 1987; Bratton and van de Walle 1997; Jinadu 2011). During the 1999, 2003, and 2007 elections, domes- tic and international observers documented many instances of overt collusion between INEC regional electoral commissioners and leading political candidates to rig elections by stuffing ballot boxes, changing election results, and disqualifying opponents (TMG 2003, 2007; see also reports by EU EOM 2003; 2007). Although reforms were made during the 2003 and 2007 election periods, it is possible that they were mere concessions that were incapable of effectively reducing the incumbent’s ability to use EMBs for his political advantage. The 2011 election period marked a significant departure from the previous 10 years of inadequate and poorly implemented EMB reforms. The first reason for this period of rapid change was the crisis of domestic and international legitimacy that followed the fraudulent 2007 elections. International and domestic observers described the elections as the most fraudulent, 71 According to Obi (2008:81) flawed elections have not affected the international standing of Nigeria and confirms that “the actions of the international community towards ‘flawed elections’ are still ambivalent, inconsistent and heavily moderated by economic, and strategic interests.” 81 violence-ridden, and poorly administered elections in Nigeria’s history (NDI 2007; JDPC 2007). In essence, the 2007 elections represented a critical juncture that provided an opportunity for meaningful policy change. Considerable blame was placed at the feet of INEC for its involvement in the rigging the election for the ruling PDP presidential candidate and widespread administrative failures, including a flawed voter register (Committee 2008; Jinadu 2011). In 2007, following his inauguration, the newly elected president, Umaru Musa Yar’Adua, initiated a comprehensive electoral reform process that eventually resulted in the first constitutional amendment of the 1999 constitution, which, among other things, increased the institutional and financial autonomy of INEC (See Table 3.4 for details). Furthermore, the sitting EMB chairperson–Professor Maurice Iwu–was replaced by a nationally reputable and non-partisan chairperson and Nigeria’s National Assembly increased its financial support for INEC to the tune of US$ 580 million (Economist 2011). The second reason for the policy changes was the commitment of the leaders of the PDP to enact meaningful electoral reform and promote a free, fair, and peaceful election (Lewis 2011). Many leading members of the ruling party, including the then President Yar’Adua and his successor President Goodluck Jonathan, believed that comprehensive electoral reform was the only way that Nigeria could reclaim its international image and revive the domestic legitimacy that had been lost in the 2007 elections (Lewis 2011). 72 A third reason for the success of the reforms in the 2011 period was that the main opposition parties gained greater strength and cohesiveness. Through the overturn of fraudulent 72 Similar sentiments were articulated through private interviews with leading PDP representatives in the National Assembly. In particular, a leading PDP Senator shared how the Senate was committed to passing legislation that improved the political and financial independence of INEC and created mechanisms to hold the political elite accountable (Personal interview, Abuja, Nigeria, June 2010). 82 election results, opposition parties gained control of the governorship in important states and picked up additional seats in the National Assembly (Jinadu 2011). The main opposition par- ties, especially the Action Congress of Nigeria and the Congress for Progressive Change (CPC), demonstrated greater unity in advancing the constitutional reform process. For instance, during early 2010 when it seemed that the reform process had been stalled, opposition parties in conjunction with leading civil society organizations, such as the Nigerian Labor Congress and the Save Nigeria Group, mobilized public protests in major capital Nigerian cities, including the capital Abuja and Lagos (LeVan and Ukata (2012)). These protests provided a groundswell of popular support for meaningful electoral reform and sent a clear message to the incumbent of the risks citizens would take to ensure that these reforms were implemented. 73 Finally, following the 2007 sham elections, the international community also adopted a different diplomatic posture towards Nigeria. Although there were no direct sanctions placed on Nigeria, there were a number of incidents that arguably highlighted that the international community, especially the US and EU, had been diplomatically distancing themselves from Africa’s most populous state. 74 Furthermore, international donors, such as the United States and the European Union, made it clear that their support for democracy programs in Nigeria was 73 On March 10th 2010, thousands of Nigerians marched through the capital of Abuja bearing placards that indicated their disgust with the country’s electoral management body, the Independent National Electoral Commission (INEC). 74 This perspective was developed through interviews with executive officials of various Nigerian political parties and executives of international donor agencies. These interlocutors generally believed that President Obama made a strategic decision to avoid visiting Nigeria during his extensive 2009 tour of Africa because of the fraudulent conduct of the 2007 elections. 83 contingent on the passage of electoral reforms and the removal of the electoral commissioner, Maurice Iwu, who presided over the 2007 elections. 75 3.4.4 Zimbabwe (2000-2008): Significant EMB Reforms No Match for the Longstanding Incumbent Zimbabwe provides the most robust test of the proposed theory. The analysis of the Zimbabwean Electoral Administration spans the 2000 to the 2008 election periods. 76 During this time there are two distinct trends in the frequency and nature of reforms in EMB autonomy and capacity. The 2000 presidential and 2002 parliamentary elections experienced no major changes in the administrative structure that had been in place since the 1996 presidential elections (Lodge, 77 Kadima and Pottie 2002). However, as highlighted in the introduction to this chapter, between the 2005 and 2008 election periods, there were fundamental changes from a mixed model of electoral administration to an independent model in which the EMB was consolidated and legally empowered to manage multiple dimensions of the election process (ZESN 2005). 78 Nonetheless, similar to Nigeria (1999-2007), President Mugabe found ways of preventing the implementation of the reforms either by appointing close associates as EMB executives or passing emergency laws that counteracted many of these reforms. What explains the rapid implementation of 75 Personal interviews with executives from international donor agencies provided support for this contention. One executive explained how funding and support for the 2011 elections were directly tied to the removal of Professor Iwu as chairman of the INEC. 76 See Table 3.11 for a list of the EMB reforms in Zimbabwe 2000-2008. 77 In fact, during this time President Mugabe introduce legislation and procedural changes to increase his grip on electoral administration (see Makumbe 2002: 95). 78 Amendment 17 of the Constitution of Zimbabwe established the Zimbabwe Electoral Commission and abolished the Electoral Supervisory Committee. 84 reforms over this eight-year period? I suggest that the proposed theory can shed some light on the President Mugabe’s willingness to formally cede control of electoral management. Let us begin the inquiry by examining the 2000 and 2002 election periods. During the 2000 constitutional referendum and the parliamentary polls, the executive had full reign over the election administrative apparatus (Lodge et al. 2002: 449). The 1979 constitution and the 1992 Electoral Act informed the four separate institutions that were responsible for election governance in Zimbabwe (Pottie 2002; Wall, for Democracy and Assistance. 2006): 79 1. The Delimitation Commission: responsible for boundary delimitation; 2. The Electoral Supervisory Commission: responsible for supervising the elections and accrediting election observers; 3. The Registrar-General of Elections: responsible for registering voters and maintaining the voter register; 4. The Election Directorate: responsible for conducting the polls. The executives of all four bodies were appointed by President Mugabe and by many accounts, were his close partisan, military and intelligence affiliates (Lodge et al. 2002: 447-9). In the 2002 presidential elections that followed, the incumbent utilized his control of the parliament to push through highly controversial electoral “reforms” in 2002 that undermined citizens’ human rights and drastically abridged the autonomy and capacity of electoral administration. Although the electoral administrative design remained the same, various laws were passed that limited the types of individuals that could monitor elections and made it impossible for Zimbabweans with dual citizenship and those living outside of the country to vote 79 See also Constitution of Zimbabwe (1979 as amended) and the Electoral Act (7/1990, 7/1992, 22/1992). 85 (Makumbe 2002). The electoral management authorities reallocated 5,600 polling stations from urban to rural areas in an attempt to maximize President Mugabe’s support in the rural regions (Pottie 2003). There are three reasons why President Mugabe and the Zanu-PF were successful in maintaining full control of the electoral administrative framework. From the perspective of the domestic balance of power, President Mugabe enjoyed strong executive powers, and possessed an extensive coercive state apparatus and loyal paramilitary groups (including the war veterans) that he used to intimidate voters and frustrate the burgeoning opposition movement (Levitsky and Way 2010; 242; Masunungure and Shumba 2012). He also wielded control of key state institutions through a network of Zanu-PF officials who were staunchly loyal to him. For instance, it is alleged that the Director of the Registrars’ General of Elections, Tobaiwa Mudede, whose department is responsible for compiling the voters’ register, is a close associate of President Mugabe (Johnson 2011). On the other hand, Zimbabwe’s opposition had only recently coalesced into a major movement during the late 1999 to advance the democratic and economic reforms, and challenge the incumbent-led overhaul of the constitution (Raftopoulos and Eppel 2008). While the main opposition, Movement for Democratic Change (MDC), had momentary success during the constitutional referendum and the 2000 parliamentary elections, they were no match for the Zanu-PF’s coercive capacity as key MDC leaders, including Morgan Tsvangirai were jailed and some were even tortured following the 2002 elections (Laakso 2002). As a result, the opposition could not effectively raise public awareness, mobilize popular support, or advance legislation in the parliament. 86 The third reason for the incumbents’ relative strength was absence of coordinated and sustained pressure from the international community. It was only in the lead up to the 2002 elections that the international community began to tighten the screws on the regime by enacting sanctions or threaten to sanction the government unless changes were made. However, there was lack of consensus among other African countries on whether sanctions were the best way to promote democratization in Zimbabwe. 80 For instance, during the 2000 elections South Africa refused to implement sanctions and many believe that this provided the Zimbabwean government with an opportunity to vacillate on reforms (Levitsky and Way 2010: 243). The 2002 presidential elections provided the critical juncture for the massive reforms in election administration that followed. Amidst the concerted pressure from the opposition and major international actors in late 2004, months before the next parliamentary elections, President Mugabe introduced massive reforms of the electoral administrative structure that sought to increase the credibility of elections by developing new institutions with greater autonomy from executive control. This was achieved through the creation of the Zimbabwean Electoral Commission (ZEC) to replace the ESC. The new EMB was responsible for supervising elections, registering voters, conducting voter education, and accrediting observers. What made the body more independent was the appointment process in which the Judicial Service Commission and 80 To be clear, the UK placed an arms embargo on Zimbabwe in 2000 and the government also froze assets and placed travel bans on leading Zanu-PF officials. The United States enacted similar sanctions to the UK under the Zimbabwe Democracy and Economic Recovery Act of 2001, but these sanctions were only implemented in 2003. The European Union placed an extensive ban on military weapons in the lead up to the 2002 elections because the government banned the EU observer mission from observing the 2002 presidential elections. Other countries such as Australia, New Zealand, and Switzerland placed arms embargoes in 2002. (Smith-Hohn 2010: 2). 87 the Parliament (through the committee on Standing Orders) had greater latitude in nominating members of the commission. 81 There were two main factors that underscored the adoption of the legislation in 2004. The first was that leaders of Southern African countries placed more pressure on President Mugabe and the Zanu-PF-led administration. Specifically, the Southern African Development Community (SADC), 82 introduced the Protocol on Principles and Guidelines Governing Democratic Elections of August 2004 that established rules of electoral administration that all countries in 83 the SADC region were compelled to institute. Along with the pressure to comply with the new SADC regulations, Mugabe experienced sharp criticism from members of the African Union (AU), pressure from South African civil society and business interests, and the ongoing sanctions from other western countries (Levitsky and Way 2010: 244). 84 While most western donors maintained and even expanded the scope of their sanctions, some donor countries including the United States and the United Kingdom, took specific steps to directly fund the MDC and key civil society groups involved in the pro-democracy movement. 81 82 83 See Amendment 17 of the Zimbabwe Constitution. SADC is a regional organization comprised of 15 countries including Zimbabwe. SADC Principles and Guidelines Governing Democratic Elections were adopted by the SADC Summit, Mauritius, August 2004. Among the main principles the new regulations mandated 1) “Independence of the Judiciary and impartiality of the electoral institutions” and 2) “Acceptance and respect of the election results by political parties proclaimed to have been free and fair by the competent National Electoral Authorities in accordance with the law of the land” SADC (2004). 84 According to the International Crisis Group (ICG), it was the combination of these various sources of domestic and international pressures that forced the hand of Mugabe: “On 7 September 2004, against the backdrop of domestic and international pressure on it to abide by principles and guidelines governing democratic elections adopted the previous month in Mauritius by SADC, the government introduced a bill to create the independent Zimbabwe Electoral Commission (ZEC) (ICG 2005: 9). 88 Equally instrumental to the reform agenda was the MDC’s increasing popularity among Zimbabweans. But, most importantly, was MDC’s threat to boycott the upcoming elections if fundamental reforms were not implemented. According to a report by International Crisis Group (ICG), MDC internal debate has intensified in the wake of the August 2004 declaration that the party would not participate in the forthcoming byelections and would boycott the March 2005 elections as well, unless ZANUPF implemented in full the SADC electoral principles and guidelines (ICG 2004: 11). As the 2008 elections approached, the Parliament passed a series of statutes and approved the 18th amendment of the Constitution, which among many things enhanced the autonomy and capacity of the ZEC (Kadima and Booysen 2009). For instance, the new electoral law guaranteed the posting of election results outside polling stations. ZEC was also given the powers of maintaining copies of the voters roll, monitoring the media supervising the multi-party liaison committee, and delimitating constituency boundaries. 85 Similar to the 2005 elections the successful passage of these reforms was predicated on the strength and activism of the opposition MDC (even though the party lost seats during the 2005 elections and split into two factions) and the increasingly active engagement of the international community, spearheaded by SADC. The international community was essential in supporting the MDC and other civil society groups such as Zimbabwe Electoral Support Network that were instrumental not only in advancing the reform agenda but also in mobilizing domestic election observers to conduct a parallel vote tabulation. Unfortunately, Mugabe still found ways of undermining the power of EMB through 85 See the 2007 Electoral Amendment Act and the 18th Amendment of the Zimbabwe Constitution. 89 the presidents’ ability to amend electoral laws without parliamentary approval and appointing Zanu-PF loyalists to key positions within the ZEC administrative structure (ZESN 2008). The 2008 Harmonized (Presidential, Parliament, and Local Government) elections were held on the 29th of March. For many domestic and international observers the elections were peaceful and competently administered by ZEC (EISA 2008; ZESN 2008). In contrast to other previous elections, many of the reforms that opposition and civil society groups had fought to be passed, such as the posting of results at polling stations and the expansion of polling stations, were for the most part implemented on election day (EISA 2008). In fact, the posting of results at each polling station allowed the opposition and civil society groups such as ZESN to conduct a precise tabulation of votes for the presidential election (ibid). Despite these reforms, ZEC took approximately 5 weeks to announce the final results for the presidential elections in clear contravention of the existing legal framework. ZEC in its official elections report blamed the delay on difficulties in accessing the source documents from the provincial collation centers and the need to conduct multiple recounts in 23 constituencies (ZEC 2008). But this perspective was widely discredited by the MDC-T and members of the pro-democracy movement. According to Derek Matyszak a leading Zimbabwean scholar, ZEC officials acted on the “instruction of persons within the ZANU-PF hierarchy, and, more particularly the military, which was reported by some as having taken over the management of the electoral process” (2009). When ZEC finally announced the results of Presidential election on 2 May 2008, neither President Mugabe or MDC-T Presidential candidate Morgan Tsvangirai received a 50 percent + 1 majority needed to win (Booysen 2009). 86 As a consequence, Zimbabweans had to return to the polls a second time to select a president. However, President Mugabe was determined to 86 Specifically, according to ZEC official results Morgan Tsvangirai received 47.9% and Robert Mugabe 43.2%. 90 prevent Zimbabweans from expressing their democratic choice in another election. Instead, Mugabe unleashed a nation-wide military-led operation, called “Votera Papi,” to find, intimidate and violently punish MDC officials and supporters (Bratton and Masunungure 2008; EISA 2008). With concern for the safety his supporters, Morgan Tsvangirai decided to boycott the presidential run-off (Booysen 2009). 3.5 Discussion In what ways does the theory of EMB reform apply to the four countries and 18 election periods under consideration? This section provides a brief comparative analysis of the different factors that contributed to the presence or absence of electoral administrative reform. In Table 3.4, I present a summary of the main dimensions of the comparative analysis. Rules that enhanced EMB autonomy were introduced in 11 of 18 election periods (61%). Of these 11 instances of rule changes the majority were major changes to EMB autonomy (Major changes= 7; Minor changes = 4). The most frequent reforms to EMB autonomy took place in Nigeria (4 election periods) and Zimbabwe (3 election periods). Turning now to capacity-related reforms. Of the 18 election periods examined in the analyses, the capacity of EMBs was enhanced in 15 election periods (83%). Major reforms were passed in 7 election periods, moderate reforms in 7 and minor reforms in 4. The data show clearly that reforms in EMB capacity occur more frequently than autonomy-related reforms. One may conclude that the higher frequency of capacity-enhancing reforms is related to: 1) greater willingness of governments to accede to demands for improving the organizational efficiency of the EMB; 2) relative ease with which improvements to capacity can be implemented through procedural changes instead of major legislative reform; 3) and extensive international funding provided to enhance EMB capacity. It is also possible that once major 91 stakeholders perceive the EMB as autonomous— as in the case of South Africa and Ghana— there is less demand and supply of reform to enhance EMB autonomy. A second finding is that the domestic political balance of power between the political opposition and the incumbent is crucial to explaining the likelihood of policy innovations in EMB autonomy and capacity (as well as the extent to which rule changes are enforced). In Ghana (1996), Zimbabwe (2005), and to some extent Nigeria (2011), opposition parties were able to pressure the incumbent into implementing important changes to the electoral administration either through boycotts, 87 threats of boycotts, or the staging of mass protests. Although the MDC was less successful in guaranteeing credible enforcement, the ability of the opposition to influence the reform agenda is in contrast to Nigeria (1999-2007) where the opposition was less coordinated and less powerful relative to the incumbent PDP. The third set of findings concern the impact of the international community. International sanctions were only applied in Zimbabwe (2002 and 2008). However, there were other instances where the loss of international legitimacy may have influenced incumbents’ commitment to reform, such as in Ghana (1996). Although sanctions such as the targeted ones place on President Mugabe and members of his inner circle (2002-2008) were effective in bringing about change, there were other less punitive forms of international pressure or ‘soft diplomacy’ that are difficult to incorporate in the analysis and might be at work in countries such as Nigeria (2011). 87 Although being identified as an important causal factor for reform, actual opposition boycotts occurred in only 4 of the 18 cases. However, in all three cases the oppositional withdrawals from the electoral contests were correlated with electoral administrative reform. The case of Ghana in 1992 provides the most striking example of this influence, but also highlights some of the other factors such as oppositional strength and concomitant international pressure that are needed for meaningful reform to occur. A more difficult factor to assess is the threat of boycotts directly related to demands for electoral administrative reform. According to my analysis these threats occurred in three cases. In Zimbabwe these threats were made explicit and preceded the reforms of the electoral administration in 2004 and 2007. 92 The impact of the international community was not only felt through their utilization of “sticks”, but through financial and technical assistance in 1) building the capacity of EMBs (this occurred in almost all the cases to varying degrees), 2) building the capacity of oppositional parties and domestic civil society organizations (this was most prevalent in South Africa (1994); Zimbabwe (2002-2008), Nigeria (2003-2011). In all the cases expect South Africa in the 2004 elections and Zimbabwe in the 2000 elections, my results indicate that international assistance was provided in some form and this support could have tipped the balance in favor of the major agents of change. Fourth, the findings indicate that rampant electoral fraud is an important catalyst for reform of EMBs. Fraudulent elections represent a critical juncture that provides greater leverage for agents of reform to achieve changes in the prevailing rules. Table 3.4 shows that important adjustments in autonomy and capacity rules occurred after fraudulent elections. I find that in the 8 cases of extreme electoral fraud and manipulation where the EMB was widely perceived as being complicit autonomy and capacity related reforms occurred approximately all cases. 3.6 Conclusion In this chapter, I have suggested that we can explain the design of EMBs in Africa by examining the strategic interaction among domestic elites and international actors. Through a comparative assessment of EMB reform in Ghana, Nigeria, South Africa and Zimbabwe, I found that EMBs are likely to have greater levels of formal autonomy when opposition parties and the international donor community exert credible and enduring pressure on the incumbent regime. In some countries, such as Ghana and South Africa, reforms have been institutionalized while in Nigeria and Zimbabwe institutionalization is ongoing. In the former examples, reforms of EMB autonomy occurred less frequently if at all as political elites have confidence in the institutions’ 93 ability to effectively manage elections. In the latter two cases, political elites are still engaged in battles to define the institutions’ autonomy. In the next four chapters (4-7), I explore the impact of EMB design and performance on electoral integrity and democratic consolidation. I suggest that there is a direct link between the analysis of the causes and consequences of EMB reforms. The processes, actors and institutional contexts that influence EMB design have direct implications for the distribution of power over election administration. Therefore, in political contexts where there are minimal safeguards to prevent executive abuse of election administration, how the EMB is designed will influence the actual and perceived performance of the institution. This in turn shapes elites’ and citizens’ electoral attitudes and behavior. 94 APPENDIX 95 Table 3.1: Reform Choices to Enhance EMB Autonomy and Capacity EMB Capacity Change EMB Autonomy No change Change (1) Credible commitment to reduce EMB manipulation (3) Non-credible commitment that may give incumbent more power to manipulate 96 No change (2) Less credible commitment that can be undermined due to weak capacity. (4) Maintain Status Quo Table 3.2: Distribution of EMB Autonomy and Capacity by Election Periods EMB Capacity EMB Autonomy Moderate/ High Low Low Ghana (1996) South Africa (1994) Ghana (1992) Nigeria (1999, 2003, 2007) 97 Moderate/High South Africa (1999 -2009) Ghana (2000-2008) Nigeria (2011) Zimbabwe (2000-2008) Table 3.3: Dimensions of Autonomy and Capacity Dimensions Autonomy Dimensions Institutional Autonomy Professional Autonomy Partisan Autonomy Financial Autonomy Capacity Dimensions Professional Capacity Administrative Capacity Election Processes Capacity Stakeholder Cooperation and Conflict Resolution Capacity Legislative and Sanctioning Capacity Indicators Institutional model of EMB Legal basis of institutional independence Appointment process of EMB commissioners Length of tenure of EMB commissioners Removal Process of EMB commissioners Whether the commission is composed of partisan or expert membership EMB financial control over operational budget Qualifications of EMB commissioners Whether the EMB secretariat is a permanent body or constituted for elections Voter registration procedures Polling procedures Counting and tabulation procedures System of communicating with political parties during elections Main EMB has the power to sanction electoral offenders Main EMB has the power to propose electoral legislation 98 Table 3.4: EMB Autonomy and Capacity Reforms and Main Causal Factors by Election Period Country/ Election Ghana 1992 1996 2000 2004 2008 2012 Autonomy Capacity Coercive Opp. Opp. Opp. Int’l Int’l Int’l Previous Transition Capacity Boycott Protests Strength Sanctions Assist Leverage Election to Fraud Democracy Moderate Major No No No No Moderate Major Moderate Moderate Moderate Moderate High High Low Low Low Low Yes No No No No No Yes No Yes No No No Low Medium High High High High No No No No No No EMB High EMB High EMB High EMB High EMB High Missin High g Yes No No No No No Yes No No No No No 1999 2003 2007 2011 South Africa Moderate Major Moderate Major Moderate Major Moderate Major Medium Medium Medium Medium Yes Yes No No Yes No Yes Yes Medium Low Low Medium No No No No/Yes EMB Low Missin Low EMB Low EMB Low Yes Yes Yes Yes Yes No No No 1994 1999 2004 2009 Major Major No No Major Major Minor Minor Low Low Low Low No No No No No No No No High Low Low Low No No No No EMB EMB EMB No Low Low Low Low No No No No Yes No No No Zimbabwe 2000 2002 2005 2008 No No Major Major No No Major Major High High High High No No Threat Yes Yes Yes Yes Yes Medium High Medium High No Yes Yes Yes OPP OPP OPP OPP High High High High No Yes Yes Yes No No No No Nigeria 99 Table 3.5: EMB Autonomy and Capacity Reforms in Ghana (1992-2012) Election Autonomy Reform Capacity Reform Period Constitutional Statutory Procedural/Regu Constitutional latory 1992 Presidential Election Laws of 1992 PNDC Law 271, 1991 Established Powers of INEC 1996 1992 Electoral 1992 Constitution Commission Act Constitution (Articles 43): 1993 (Act 451) (Articles 44): Establishes EC Functions details Register Voters; composition, Delimit terms of office Constituencies, and Results appointment announced at the (Article 46) polling station Constitutional (Article 45:13), independence Power to (Article 55) Sanction Power to make Electoral regulations Offenders (Article 54) Financial Independence 100 Statutory Electoral Commission Act 1993 (Act 451) Public Elections Regulations, 1996 (CI15) Voter ID cards Procedural/Regu latory Counting at Poling Stations Local announcement of results by presiding office Public Elections Regulation of Voters Regulations 1995 New voter Registration; Voters ID card with photo; Counting at polling stations; Inter Party Advisory Committee, (IPAC Code of Conduct Table 3.5 (cont’d) 2000 Political Parties Law (Act 574) of 2000: Allowed EC to set limits on party financing 2004 2008 101 Voter ID cards w/photo Voter Registration based on optical mark readers (OMR) The installation of a WAN (Wide Area Network) to link the regional and district offices to the head office Code of Conduct Voter ID cards with two pictures on the voters roll Faxing the results to the general office Regional scoreboards were eventually installed Table 3.5 (cont’d) 2012 Public Elections (Registration of Voters) Regulations 2012 CI 72 102 Electoral Commission of Ghana 2012 Framework for Domestic Elections Observation Table 3.6: EMB Autonomy and Capacity Reforms in South Africa (1994-2009) Election Autonomy Reform Capacity Reform Period Constitutional Statutory Procedural/Reg Constitutional ulatory 1994 Interim 1993 Electoral Constitution Commissions Act 1993 1999 2004 Constitution of the Republic of SA (1996): constitutional autonomy of commission Electoral Commissions Act, 51 1996: Establishes role and function of IEC: new appointment process; tenure for EMB commissioners; appointment of chief electoral officer Constitution of the Republic of SA (1996): Ensure free and fair elections and announce results in shortest possible time Independent auditors to monitor the results process 103 Statutory Electoral Commissions Act, 51 1996: Establishes role and function of IEC Electoral Act 1998: Conflict Management Committees; Political Party Code of Conduct; Voters roll; Electoral Laws Amendment Act, 2004: Voting in any province outside of registration district Procedural/Reg ulatory New voter registration system(“ZipZip”) machines; results verification system; counting at polling stations; increase in the # of polling stations Increase # of polling stations; voter registration verification online and SMS; Results Tabulation Center; Voting Table 3.6 (cont’d) station monitoring; new voter education campaign Increase the number of voting stations; student registration drives; scanning results slips to reduce errors with counting results 2009 104 Table 3.7: EMB Autonomy and Capacity Reforms in Nigeria (1999-2012) Election Period Autonomy Reform Constitutional 1999 2003 2007 1999 Constitution: Establishment of INEC: Powers, functions, composition of INEC Capacity Reform Statutory Procedural/Reg ulatory Statutory Procedural/Reg ulatory Electoral Act 2003: INEC Responsible for compiling Voter Register; the counting takes place at the polling stations Electoral Act 2006: New procedures for voter registration; Responsibility to promote voter and civic education; Power to prosecute electoral offenders Decree N. 17 of 5 August 1988 Electoral Act 2003 Constitutional Computerized Voter Register using Optical Mark Recognition Distribution of VR Cards 1999 Constitution Electoral Act 2006: Increased Financial Autonomy; INEC appoints INEC Secretariat instead of executive 105 Electronic Voter Registration based on Direct Data Capture machines Table 3.7 (cont’d) 2011 2010 Amendment to the 1999 Constitution Electoral Act 2011 Use of credible members of society to announce results INEC sets the date and order of elections; financial autonomy of INEC Bars political representatives from being appointed as INEC commissioners 106 2010 Amendment to the 1999 Constitution: INEC authority to promote internal-party democracy Electoral Act 2011: Posting election results at polling stations Modified Open Ballot System of Voting; National Youth Service Corps as temporary polling staff; Table 3.8: EMB Autonomy and Capacity Reforms in Zimbabwe (2000-2008) Electio Autonomy Reform Capacity Reform n Period Constitutional Statutory Procedural/Regu Constitutional latory 2000 2002 2005 Constitutional Amend. 17 (1990): Outlines the functions of ZEC abolishes ESC Statutory Citizenship of Zimbabwe Amendment Act 2002 Electoral Amend, Act 2002: Additional powers to Registrar general to alter voter register ESC exclusive power to accredit domestic and international observers Constitutional Amendment 17: Revision of eligible voters Zimbabwe Electoral Commission Act, 2005 107 Procedural/Regu latory Computerized voter register Table 3.8 (cont’d) 2008 Constitutional Amendment 18((1990): Electoral Amendment Act, 2007: Justice Minster approves ZEC regulations Constitutional Amendment 17: ZEC responsible for delimitation Zimbabwe Electoral Commissions Act, 2008 108 Electoral Amendment Act, 2007: counting at poling station; ZEC keeps copy of voters roll; monitor the media; Foreign Ministry can veto votes Increase # of polling stations; results posted at polling station CHAPTER 4 THE CONSEQUENCES OF EMB DESIGN ON ELITES AND CITIZENS IN AFRICA 4.1 Introduction In the next four chapters (4, 5, 6 and 7), I shift the focus of analysis from explaining variations in the autonomy and capacity of electoral management bodies (EMBs) to understanding the effects of EMBs on the quality of elections in Africa. In other words, the democratic quality of elections becomes the object of explanation (dependent variable) and I examine the extent to which differences in the autonomy and capacity of EMBs (independent variables) may help explain why some elections are free and fair when others are not. As mentioned in the Chapter 1, I expect that EMBs matter for both elites and citizens in emerging democracies because the performance and design of the institutions directly influence the likelihood of electoral manipulation and the incidence of electoral administrative irregularities. In Figure 4.1 I replicate Figure 1.2 displayed in Chapter 1 that outlines the proposed relationships among EMB design, EMB performance, and citizens’ and elites perspectives on election quality. I argue that citizen and elite perspectives on elections are shaped by (1) EMB design, (2) EMB performance and, (3) the overall political and institutional 1 context. I define “EMB design” as the formal rules that define the function and role of the electoral management body. EMB performance reflects the extent to which the formal rules are enforced and effectively implemented, the impact of informal rules and other aspects of the prevailing political context, including the rule of law, level of democracy, and institutionalization 1 All figures and tables are placed in the Appendix at the end of each chapter. 109 of other state and political institutions. I also distinguish between EMB autonomy and capacity and apply this conceptual distinction to both EMB design (de jure autonomy/capacity) and EMB performance (de facto autonomy/capacity). While both EMB design and EMB performance are important, I argue that in the context of emerging democracies EMB performance should more effectively explain variation in citizens and elites’ electoral attitudes and behavior. More specifically, this chapter examines the first part of the model described in Figure 4.1: the relationship between EMB design (de jure autonomy/capacity) and citizens’ and elites’ perspectives of election quality (numbered 1A and 1B). The chapter is guided by two primary research questions: 1. Does the design of EMBs (de jure autonomy and capacity) influence elite perspectives on election quality? 2. Does the design of EMBs (de jure autonomy and capacity) influence citizen perspectives of election quality? Using historical institutionalism as a theoretical frame, I suggest that variations in EMB design (de jure autonomy and capacity) may shape the strategic interaction of elites as well as their evaluations of electoral integrity. When EMBs enjoy high levels of autonomy and capacity, the incumbent—and political elites more generally—find it more difficult to manipulate the EMB and engage in more general acts of electoral fraud. Moreover, EMB officials are better able to resolve the challenges of election administration and in so doing, reduce the magnitude of electoral administrative irregularities. All things being equal, I expect EMB autonomy and capacity to have positive implications for the strategic interaction between elites and the overall quality of elections. In particular, political elites are more likely to play by the rules of the game (and eschew fraud and violence), participate in elections, and accept electoral defeat without 110 crying foul. However, there are limitations on the effectiveness of EMB design in shaping elite behavior, mainly because incumbents often resort to informal politics to prevent the institutionalization of EMBs, as was seen with the examples of Presidents Mugabe of Zimbabwe and Obasanjo of Nigeria highlighted in Chapter 3. At the same time, I expect citizens to respond to variations in EMB design in two primary ways. First, and more directly, EMB design can signal the incumbents’ commitment to making the playing field level and reducing the incidence of electoral irregularities. A second and more indirect effect is that citizens respond to the strategic interaction of elites. The extent to which elites play by the rules of the game, participate in elections, and accept electoral defeat have implications for how citizens view the quality of elections. Similarly, for citizens the influence of formal rules may be imperfect because accessing information about the complexities of EMB design may be costly for everyday citizens, and limited to those who are well educated and interested in politics. In sum, the impact of EMB design only tells a part of the story, especially in emerging democracies. This is because political actors are more likely to respond to the actual performance of EMBs that is determined by the extent to which formal rules are enforced as well as the characteristics of the political and institutional context. Nonetheless focusing on the potential influence of formal rules is important for the following reasons: First, from a theoretical perspective, even though institutionalization of formal rules related to EMB design may not occur immediately, these rules and procedures have important con- sequences for the incentives and behavior of political actors. For instance, as I highlighted in Chapter 3, political elites are constantly battling over the rules that define electoral management, which indicates that the rules have some degree of significance. Moreover, formal 111 rules may have greater effect in some countries because of the prevailing political institutional con- text, such as the political commitment to free and fair elections, and the presence of institutions that can check the power of the executive. Also, it might be the case that some dimensions of EMB design matter more than others in shaping the preferences and behaviors of political actors. Second, from an empirical perspective, the rules and procedures that explain variations in EMB design are easier to measure and use in order to make comparisons across countries and within countries over time (when compared to those that measure EMB performance). Third, if formal rules do not matter or matter little, this has important implications for policymakers who put faith in institutional engineering as the main strategy for promoting democratization. I empirically assess the consequences of EMB design (de jure autonomy and capacity) for 1) elite perceptions of election quality, and 2) citizen perceptions of election quality. To conduct these analyses, I utilize original data on various indicators of de jure autonomy and capacity in 20 African countries. I first examine the effect of EMB design on elites using an overall mea- sure of election quality derived from Staffan Lindberg’s Elections in Africa Database (Lindberg 2006). The main findings reveal that my composite indicators of EMB de jure autonomy and capacity are not significantly associated with elite perceptions of election quality. I then examine the effect of EMB design (independent variable) on citizens’ election quality perceptions (dependent variable). To gauge citizens’ attitudes towards election quality, I use recent public opinion data from the Afrobarometer across 18 countries between 1999 and 112 2008. Similar to the findings for elites, citizen perceptions of election quality in Africa are not significantly associated with either composite index of EMB autonomy or capacity. The absence of a consistent relationship between EMB de jure autonomy and capacity and political attitudes has important implications for how we understand the role of EMB design in Africa. We may infer from these results that in many countries EMB reforms are not being adequately implemented, or that EMB reforms on their own are not effective in altering elite and citizen attitudes in countries with lower levels of democracy and underdeveloped political institutions. This has crucial ramifications for the policy community. In particular, it high- lights the challenges associated with relying only on institutional engineering and the need for greater focus on establishing effective electoral management through a network of democratic institutions. The following section of the chapter briefly explores the existing literature on the effect of EMB design on political behavior and outlines the main hypotheses linking EMBs to elites and citizen electoral attitudes. I then test applicability of theory for elites and citizens in Africa, and conclude with a brief discussion of the main findings. 4.2 Literature Review Scholars have identified two broad categories of institutional rules that structure elites’ 2 and citizens’ perceptions of election quality. The first, and most widely studied, are the rules of electoral competition such as electoral formula, district magnitude, electoral franchise, and the 2 According to Mozaffar and Schedler (2002) there are three levels of electoral governance: rule-making, 2) rule application, and 3) rule adjudication. At the rule-making level there are two basic components: 1) rules that structure electoral competition, such as electoral formula and district magnitude and, 2) rules that shape electoral governance, such as campaign financing, voting, counting and tabulation, and election management institutions, etc. 113 electoral timetable. 3 The second category includes rules that shape electoral governance 4 (electoral administration). These rules encompass the design of electoral management bodies (electoral commissions), dispute resolution organizations (electoral courts) and processes such as voter registration, counting and tabulation of votes, and election observation. As indicated, in Chapter 1, I focus primarily on the literature that examines the influence of electoral administrative rules on citizen and elite perceptions of election quality. 4.2.1 Effect of EMB Autonomy on Elite Attitudes In recent years, scholars have become increasingly interested in the relationship between the design of EMBs and mass and elite perspectives on election quality (Birch 2008; Hartlyn, McCoy and Mustillo 2008; Alvarez, Hall and Llewellyn 2008). Most studies examine the consequences of EMB de jure autonomy with the general notion that de jure autonomy provides some indication of the extent to which the incumbent has control over election administration and whether the incumbent will use his/her prerogatives and rig the elections in his/her favor. Recent studies have integrated insights from the extensive literature on principal agent and delegation theory to highlight the electoral dilemma faced by political actors with responsibility for organizing elections in which they have a vested interest (Alvarez and Hall 2006). The greater political involvement in the organization of elections, including the 3 Studies exploring the effects of rules associated with electoral competition are too numerous to mention here. See the following for cross-national studies that examine the impact of electoral systems on citizens’ attitudes and behaviors: Norris (2004); Cox (1997); Bowler and Donovan (2012); Powell (2000). See also specific cross-national studies in Africa: Lindberg (2005); Cho and Bratton (2006); Vengroff and Ndiaye (1998). 4 For the most part, the legislature and transitional governments (in the case of the transitional democracies) create the rules that structure electoral competition. Rules that shape electoral governance are decided both by the government and EMBs depending on the level of autonomy and discretion that the EMB possesses. Both types of rules are either embedded in the constitution, in electoral laws, or in established conventions. 114 appointment and dismissal of EMB executives, the funding of the EMB operations, and the development of procedures including the opening time and location of polling stations, the more likely the perceived legitimacy of election administration will be called into question. To put it more simply, an executive may find it increasingly difficult to credibly commit to organizing impartial and professional election administration when the executive maintains complete control over the institution. One way of resolving these credibility problems is by delegating some or all of the responsibility for electoral management to an independent agency (Magaloni 2006b). One group of scholars has sought to explore the relationship between the de jure autonomy of EMBs and elite-based measures of electoral quality (Hartlyn, McCoy and Mustillo 2008; Rosas 2010). Jonathan Hartlyn and his co-authors (2008) conducted this seminal research 5 on the effect of EMB autonomy on election quality in Latin America. They argue, “Permanent electoral commissions are deemed especially important in situations of low administrative state capacity where there is a high level of distrust among political actors and few, if any, alternative mechanisms to help ensure honest, impartial elections” (2008:78). The main dependent variable—electoral quality—is based on the views of nonpolitical elites, including international and domestic election observers and journalists. The authors developed two main measures of EMB autonomy. The first is “formal-legal independence” which measures the appointment process and tenure of EMB commissioners as stated in law. The second is a “partisan” autonomy—which gauges the independence of the body appointing EMB com- missioners and the actual independence of the appointees. Importantly, they find that “formal- legal independence” (or de jure autonomy) does not have an impact on overall election quality, while partisan autonomy has a positive effect of election quality. 5 Data based on 104 elections in 19 countries over a 23-year period. 115 Other aggregate studies have found results that largely corroborate the findings by 6 Hartlyn et al (2008). For example, Birch (2011) finds that de jure partisan autonomy has a positive and statistically significant effect on electoral conduct, but de jure institutional 7 autonomy is not statistically significant. In a similar fashion, Rosas (2010) examines the influence of partisan and professional autonomy on elite attitudes about election quality in 18 Latin American countries. What makes his research more path breaking is the use micro-level data collected through a structured survey of elites. For Rosas, professional autonomy reflects EMB staff whose appointments are insulated from elected politicians, while partisan autonomy is 8 achieved when political parties are not involved in the appointment of EMB staff. Furthermore, Rosas finds that elite perceptions are positively correlated with professional autonomy, but 9 partisan autonomy is not. There is also a voluminous case study oriented scholarship that examines the influence of EMB rules on elite perspectives on elections and the process of democratization more generally. Most of these studies highlight the limited impact of de jure EMB autonomy on elite confidence in elections; see Debrah (2011) and Jinadu (2011) for research on Ghana, Makulilo (2009) for research on Tanzania, and Adebayo and Omotola (2007) for research on Nigeria. In theory, EMBs with greater de jure autonomy should be expected to promote elite’s confidence in the quality of elections; the existing empirical research is mixed. Some of the 6 Partisan autonomy gauges the degree of political party involvement in the membership of EMBs. 7 Specifically, Birch distinguishes between independent, mixed and government-run EMBs. 8 The measures of partisan and professional autonomy are based on work by Hartlyn et al. (2008). 9 Partisan autonomy becomes significant and negative when it is interacted with autonomy. This means that greater involvement of political parties in EMBs increases elites perceptions of election quality only in democratic contexts. 116 findings from the cross-national studies are inconsistent. Moreover, the effect of formal-legal autonomy is often contingent upon features of the political and institutional context such as the level of democracy or the respect for rule of law. These findings should not be surprising, especially given the findings in Chapter 3 where some incumbents can appease domestic opposition and the international community by improving the autonomy and capacity of EMBs on paper, but taking various steps to block the effective implementation of these reforms in practice. 4.2.2 Effect of EMB Autonomy on Citizen Attitudes Existing research also examines the effect of EMB autonomy on mass perceptions of election quality. In an article-length study, Sarah Birch explores the consequences of EMB autonomy on citizens’ perceptions of election quality across a range of developed and developing democracies using a measure of de jure institutional autonomy (2008). 10 The study by Rosas cited above also examines the effect of professional and partisan autonomy for citizens using recent survey data from the Latinobarometer (2010). Both studies introduce innovative methods for modeling the correlates of mass perceptions of election quality using cross-national data. As with the research on elites, the findings from these two studies are inconsistent. Birch finds that citizens living in countries with autonomous EMBs are more likely to have an unfavorable view of elections. She attributes this counterintuitive finding to the possibility that formal-legal independence might not reflect the actual independence of the institution. In a similar manner, Rosas reveals that EMB partisan and professional autonomy do not have a 10 In her research, Birch identifies three types of EMB models: government, independent and mixed. As highlighted in Chapter 2, this conceptualization of independence was originally purported by Lopez-Pintor (2000). 117 statistically significant relationship with citizen confidence in elections. 11 Although both studies utilize different measures of EMB autonomy and conduct their analysis on different sample of countries, there is no consensus on whether EMB autonomy is associated with perceptions of election quality. In this chapter, I seek to examine the effect of EMB de jure autonomy on elites and citizens attitudes towards elections by proposing the following hypotheses: • (H1a) When EMBs possess greater de jure autonomy elites are more likely to consider elections free and fair • (H2a) When EMBs possess greater de jure autonomy citizens are more likely to consider elections free and fair 4.2.3 Effect of EMB Capacity on Elites and Citizens A principal shortcoming of the cross-national literature on the consequences of EMB design is that scholars have focused disproportionately on EMB de jure autonomy, while not paying enough attention to the effect of EMB de jure capacity on popular and elite attitudes towards elections. I suggest that Africans are not only interested in knowing whether EMBs are free from political manipulation but also in whether EMBs have the necessary organizational, logistical, professional, and financial capacity to effectively execute all the complex operations associated with elections. Therefore, examining EMB capacity as a dimension of EMB design could enhance our understanding of how political actors develop their perceptions of election quality. One reason for the significance of EMB capacity is that it provides some indication of 11 Rosas shows that the effect of formal autonomy is muted when he controls for democracy; that is, citizens are more likely to express higher levels of confidence in elections within countries with higher democratic standards (2010: 85). 118 the executive’s and EMB officials’ commitment to organizing elections effectively. For instance, the introduction of a completely new voter registration system in Ghana during the 1996 elections that included citizens receiving voter registration cards signaled to the political opposition and citizens more generally the commitment of the incumbent to reduce the possibility of electoral fraud (Gyimah-Boadi 1999; Lyons 1997). As indicated in Chapter 2, there is a growing number of scholarly and policy-oriented studies that examine EMB capacity as a dimension of EMB design (Hale and Slaton 2008; Wall, for Democracy and Assistance. 2006; Lopez-Pintor 2000). By advancing this research on bureaucratic capacity they highlight some important dimensions of capacity that should more generally improve the EMBs’ ability to conduct elections. This may translate in greater confidence in the EMB and more favorable perspectives of election quality. These studies have highlighted various dimensions of EMB capacity that are essential to the quality of elections. However, only a handful of studies have systematically explored the potential consequences of variations in EMB de jure capacity for elites and citizens (Elklit and Reynolds 2000; Schaffer 2008; Alvarez, Hall and Llewellyn 2008; Tomz and Houweling 2003). 12 An example of this is seen with Elklit and Reynolds’ research on the impact of election administration on democratic legitimacy in Africa (2000). These authors demonstrate through a qualitative analysis how various dimensions of EMB capacity, such as the professional capacity of poll workers, can positively influence popular and elite opinions of election quality. 12 13 Other For more studies on EMB capacity in Africa see Alabi 2009; Ajayi 2007; Makara, 2007; Maphunye, 2010; Fall et al. 2011; Everensel et al 2010. For studies on EMB capacity in the United States see: Alvarez et al. 2008, Burden et al. 2011. 13 Elklit and Reynolds incorporate data on elections in Botswana, Burkina Faso, Ghana, Mozambique, Sierra Leone, South Africa, Tanzania, and Zambia. 119 studies have also probed the consequences of rules affecting voter registration (Schaffer 2008; Evrensel 2010), voter operations (Alvarez et al., 2008; Tomz and Houweling 2003), and counting and tabulation of results (Schaffer 2008). For example, consider the comprehensive comparative study conducted by Evrensel (2010) on voter registration systems in Africa. She shows how even well-intentioned improvements in voter registration capacity, such as the adoption of electronic voting, can have negative consequences for citizens as well as political elites when they are not effectively designed. Finally, another study by Alvarez et al. (2008) examines the effects of electronic voting systems on Americans’ perceptions of election quality. Because of the problems linked to electronic voting systems during the 2000 and 2004 U.S. elections, voters who used these advanced voting methods were less likely to believe that their votes were accurately counted. I examine the effects of EMB de jure capacity on both citizen and elite attitudes on election quality through the following two hypotheses • (H1b) When EMBs possess greater de jure capacity, elites are more likely to consider elections free and fair; • (H2b) When EMBs possess greater de jure capacity, citizens are more likely to consider elections free and fair. 4.3 Consequences of EMB Design on Elite Views of Election Quality In this section of the chapter, I explore the relationship between EMB design (independent variable) and elite attitudes on election quality (dependent variable). I suggest that when EMBs possess greater de jure autonomy and capacity, it may increase the costs for the political elite to engage in fraud and thereby reduce the likelihood of election irregularities. One 120 way in which elites may respond to the more autonomous and capable EMBs is by considering the resulting elections to be more free and fair. 4.3.1 Dependent Variable: Overall Election Quality To gauge elite perspectives of election quality, I utilize a measure from Lindberg’s Elections in Africa Database (2006). According to Lindberg, the variable captures whether electoral procedures were followed, whether people could join a political organization of their choice, whether demonstrations were allowed, and whether information was freely available. The variable is coded on a four-point scale 0-3, with 3 indicating the election was entirely free and fair. 14 Between 1990 and 2004, 25 percent of elections were judged somewhat free and fair with irregularities affecting the outcome, while 64 percent of elections had irregularities that did not affect election quality. The remaining 11 percent of elections were either “entirely free and fair” (5%) or “not free or fair” at all (4%). Admittedly, judgements about overall election quality do not directly measure the perspectives of political elites that could have been assessed through micro-level data on elites perspectives. Nonetheless, the measure reflects many of the main factors that political elites and major election stakeholders including international and domestic observer delegations would take into account when making assessments about election quality. Moreover, Lindberg’s measure of overall election quality is also strongly correlated with other more direct measures such as opposition participation, losers’ acceptance of results, and electoral turnover (Lindberg 2006a). Finally, numerous studies have incorporated measures of aggregate 14 In this four-point scale: 0 = “No” when no elections have been held or those held have been wholly unfair; 1 = “Irregular” when there were serious defects that influenced the results; 2 = “Yes, irregularities not significant” when there were deficiencies but they did not affect the result; and 3 =“Yes” when elections were free and fair. 121 election quality as a proxy for elite perspectives (Hartlyn, McCoy and Mustillo 2008; Kelley 2012). 4.3.2 Independent Variable: EMB Design To measure our main independent variable, EMB design, I introduce an original data set that measures various dimensions of EMB de jure autonomy and capacity across 19 African countries between 1990 and 2004. 15 The data set reflects information collected through content analysis of country constitutions, electoral laws, and election observation reports across the election periods included in the sample. As a first step, I construct composite indicators of EMB de jure autonomy and EMB de jure capacity. Following this, I statistically examine the relationship between the autonomy and capacity indexes and overall election quality. As before, I define autonomy as an institution’s ability to make decisions independently of the control of the executive, the political elite, and other state and society actors. I construct a composite index of EMB de jure autonomy based on six indicators of EMB autonomy: 1) institutional autonomy; 2) legal autonomy; 3) partisan autonomy; 4) the EMB commissioner appointment process; 5) EMB commissioner tenure; and 6) EMB financial autonomy. 16 • Institutional autonomy is the most commonly used de-jure measure of EMB autonomy. Institutional autonomy measures the extent to which the executive branch of government is responsible for the administration of elections. EMBs can either be government- run, 15 The nineteen countries include Benin, Botswana, Burkina Faso, Cape Verde, Ghana, Kenya, Lesotho, Madagascar, Malawi, Mali, Mozambique, Namibia, Nigeria, Senegal, South Africa, Tanzania, Uganda, Zambia, and Zimbabwe. Liberia is excluded because its most recent post-conflict of 2005 election was held after the end of the period under investigation in this research. 16 Individual coding for these variables is included in Appendix A. 122 mixed (both government and independent), and independently administered. Various scholars associate independently administered EMBs with high quality elections because the other models are often ineffective in coordinating sufficient resources to effectively organize elections. Another advantage of the independent model is that it pro- motes higher confidence among elites because the incumbent can credibly commit to not manipulating election administration (Lopez-Pintor 2000; Mozaffar 2002). • Legal autonomy indicates whether the specific model of election management is stipulated in the constitution or electoral law. I expect that when election management is constitutionally defined, political elites find it more difficult to alter key aspects of the electoral administrative framework through legislative reform (Wall, for Democracy and Assistance. 2006). • Partisan autonomy is another widely used measure of de jure EMB autonomy. It measures whether EMB commissioners are selected on the basis of their affiliations with political parties or their professional expertise. I expect that when EMBs are comprised of professionals, they are more likely to overcome the challenges of election organization, relative to commissions comprised of partisan representatives or those who combine both partisan and expert members (Hartlyn et al., 2008). • Appointment process of EMB commissioners is based on the number of politically relevant actors involved in the appointment of EMB commissioners as stated in law. Various studies indicate that the autonomy of EMBs is positively associated with the number of institutions and political actors involved in the appointment process (Birch 2008; Hartlyn et al., 2008). • Tenure length of EMB commissioners is a straightforward measure that indicates the 123 number of years EMB commissioners are appointed to their posts as well as the possibility of term renewals. I expect that in countries in which the tenure of EMB commissioners is longer than that of the executive or members of the legislature (including reappointments), the commissioner should be expected to enjoy greater insulation from political manipulation (Hartlyn et al., 2008). • Financial autonomy is a newly developed measure that gauges the level of control the EMB has over its operational budget. In this case, the fewer state and government agencies that are involved in the appropriations of EMB funding the greater the level of financial autonomy. EMBs that have a separate line in the national budget that is determined by parliament is more financially autonomous than EMBs that rely on the finance ministry or the interior department to decide on the yearly budgetary allocations (Wall, for Democracy and Assistance. 2006). To develop the composite measure of EMB de jure autonomy, I first normalize the six indicators on a 0-1 scale with 1 representing greater autonomy. Next, I conduct a factor analysis (principal components with oblique rotation) using the six EMB autonomy indicators. The results of the factor analysis are displayed in Table 4.1 that also highlights the factor weights that correspond with the various indicators. 17 Overall, the table shows that there is one underling factor in the data (Eigenvalue = 3.71; Proportion of variance = 85%). I utilize this factor to develop a composite indicator of EMB de jure autonomy (EMB autonomy index). For ease of interpretation I normalize the EMB autonomy index on a 0-1 scale with 1 indicating high levels of autonomy (mean = 0.62; standard deviation = 0.28). On one end of the EMB autonomy index 17 The values of the weights provide an indication of how much a particular indicator contributes to the overall component. 124 we find countries such as Benin (1991: 0), Cape Verde (1991: 0) and Mali (1992:0) had the lowest EMB autonomy scores (0). EMB autonomy reaches the highest levels in South Africa (1999,2004: 1.0), Ghana (1996, 2000, 2004: 0.9) and Tanzania (1995, 2000: 0.89), while moderate levels of EMB autonomy are associated with elections in Nigeria (1999: 0.64) and Malawi (2004: 0.52). The second dimension of EMB design is de jure capacity. In this respect, capacity reflects an institution’s ability to implement its decisions and legally mandated responsibilities in an effective and efficient manner. I gauge de jure capacity through a composite index constructed from five indicators: 1) professional capacity; 2) administrative capacity; 3) voter registration capacity; 4) conflict resolution capacity; and 5) capacity to sanction political actors. • Professional capacity reflects the extent to which EMB commissioners are mandated by law to have certain professional qualifications to be eligible for appointment. For instance, in Zimbabwe (2005) the chairperson of the commission must be qualified to be a judge of the Supreme Court or the High Court. EMBs comprised of commissioners with professional qualifications are more equipped to effectively plan and implement EMB operations (Birch 2007). • Administrative capacity indicates the geographic and temporal scope of the EMB administrative structures. EMB administrative capacity is enhanced when EMBs have a permanent secretariat with units in different sub-regions of the country (Wall et al., 2006). • Voter registration capacity measures the accuracy and integrity of the voter registration system as stated in law. One way of gauging this accuracy and integrity is by assessing the type of voter registration technology, which ranges from a paper-based system to a 125 computerized register based on biometric data. I suggest that computerized voter registration systems promote greater accuracy and integrity of the voters roll (Elklit and Reynolds 2005). • Conflict resolution and political-party cooperation capacity examines the extent to which the EMB has established procedures that facilitate communication and mitigation of conflict between political parties. EMBs are more likely to promote political party co- operation when they establish political party liaison committees that function as a for- mal/informal forum for the dissemination of EMB decisions and resolution of inter-party disagreements. • Sanctioning capacity assesses the range of powers that EMBs have during elections, including the power to legally sanction political parties and elites that violate electoral law. I expect that EMBs with higher levels of jurisdictional capacity will be more effective in holding political elite accountable to electoral rules and reduce elites incentive to engage in fraud (Fall et al, 2011). I create a composite index of EMB de jure capacity, in a manner similar measure of EMB de jure autonomy. The results of the factor analysis are indicated in Table 4.2 and show one underlying factor (Eigenvalue = 2.46; Proportion of variance = 96%). The corresponding EMB capacity index is normalized on a 0-1 scale with 1 indicating high levels of capacity (mean= 0.42; standard deviation 0.335). A perfect example is South Africa where EMB capacity increased from 0.327 during the 1994 elections to 1 in both the 1999 and 2004 elections (rep- resenting the country with the highest EMB capacity in our sample). 4.3.3 Controls 126 I also consider several factors that the literature identifies as being important predictors of election quality. First, I control for the effect other electoral rules, such as the proportionality of the electoral system and whether the system of government is presidential or parliamentary. Following the existing literature, I anticipate that proportional representation systems are associated with higher election quality, because proportional representation rules provide greater mechanisms for representation of minority groups relative to more majoritarian systems (Lindberg 2006; Birch 2011; Kelley 2012). Second, I account for the level of fraud and electoral violence that occurs during the election period, both of which I expect to be negatively related to election quality (Collier and Vicente 2010; Bratton 2008). 18 The third set of controls accounts for other election characteristics, such as the competitiveness of the elections, time since last elections, and whether international observers participated in the elections. Finally, I account for the effect of democracy through a lagged Freedom House index of political rights and civil liberties. 19 Accounting for the level of democracy is essential as various scholars have emphasized the relative value of EMB performance in transitional and emerging democratic regimes relative to more consolidated democracies (Pastor 1999). 4.3.4 Results 18 Including these controls raise endogeneity issues because most elite assessments incorporate the level of violence and fraud. However, I include these controls for two main reasons. First, I want to ensure that the results do not overestimate the effect of EMB design. Second, various studies have also utilized similar measures to examine the cross-national variation in election quality (Kelley 2012; Birch 2011). To ensure that the results were not significantly biased by the inclusion of these variables I ran separate regression models without fraud and violence and found similar substantive results. 19 The index is coded from 0-1 with 1 indicating high democracy. To eliminate the endogenous relationship between the Freedom House index and our measure of election quality, I create a lagged variable, meaning that the Freedom House index reflects the score for the year preceding the election because indicators in the political rights subs-score might be correlated with our measure of overall election quality. 127 Does EMB de jure autonomy and capacity help to explain variations in election quality in Africa? Before investigating the effect of EMB design on overall perceptions of election quality using multiple regression, I highlight the variation of de jure autonomy and capacity indexes across different quality elections. As summarized in Figure 4.2 there is a curvilinear relationship—an inverted “u”—between indicators of EMB design and overall election quality. The lowest average de jure autonomy and capacity are associated with elections that are not free and fair and also with those that are free and fair. Interestingly, both autonomy and capacity are highest in elections with significant irregularities. This result provides the first indication that the relationship between de jure autonomy and de jure capacity of EMBs in Africa may not be not consistent with my main hypotheses. Following this, I examine the same relationship using a multiple regression model that incorporates the control variables. I conduct an ordered logit regression model because the main dependent variable—overall election quality—is measured on an ordinal scale from 0-3. The results of the regression analysis are displayed in Table 4.3 Model 1. They indicate that both the de jure autonomy and de jure capacity indexes are not statistically associated overall election quality. We may infer from the findings that across the 19 countries included in the sample, neither measure of EMB capacity or autonomy enhances overall perceptions of election quality, after controlling for important covariates. Overall, I find that across the sample of 19 countries my composite indexes of de jure EMB autonomy and capacity are not significantly associated with overall election quality. How should we interpret these findings? From a statistical perspective, our composite measures may reflect the divergent effects of individual attributes of EMB de jure autonomy or capacity. From 128 a substantive perspective the findings highlight the lack of institutionalization of formal rules that define the function of EMBs during elections in some of the countries included in the sample. 4.4 Consequence of EMB Design on Citizen Views of Election Quality In the previous section, I found that EMB design does not systematically explain variation in election quality from the perspectives of political elites in Africa. The next step is to consider whether or not features of EMB design may have an influence on the election quality perceptions of ordinary Africans. I also suggested that citizens are more likely to consider elections as possessing democratic qualities when EMB rules provide for greater insulation from executive manipulation and when they help to reduce the difficulties associated with election management. However, in some countries citizen’ attitudes may not be influenced by the design of EMBs. This could be because EMB rules are not effectively enforced or because gaining accurate information about EMBs is too costly for ordinary citizens. To explore the hypothesized association between the perceptions of election quality of ordinary Africans (dependent variable) and EMB de jure autonomy and capacity (independent variable), I incorporate data from various sources. First, I include individual-level data from the Afrobarometer based on public-opinion surveys conducted in 18 countries between 1999 and 2008. 20 capacity. Second, I incorporate data on the characteristics of EMB de jure autonomy and 21 As a result, the scope of the analysis ranges from 1996 to 2004, based on a sample size of approximately 41,000 respondents nested in 28 election-periods (in a total of 18 20 See Table 4.4 for a description of the 28 election periods and 18 countries included in the sample. 21 See the Appendix for a detailed description of the data and variables included in the analysis. 129 countries), with an average of 1590 respondents per election period. 22 See Table 4.4 for the countries and Afrobarometer Rounds. Because the sample includes data that are measured at two levels of analysis (individual and election-period), I utilize a multilevel regression model (Steenbergen and Jones 2002). 23 Furthermore, because the main dependent variable—citizens’ election quality perceptions—is measured dichotomously, I estimate a random-intercept logistic multilevel regression model assuming that the effects of the independent variables reflect the change in the log-odds of citizen election quality perceptions (Raudenbush and Bryk 2002). 24 4.4.1 Dependent Variable: Citizens’ Election Quality Evaluations The main dependent variable—perceived election quality—captures respondents’ perceptions of the quality of the most recent elections. Respondents were asked “On the whole, how would you rate the freeness and fairness of the last national election held in XXXX?” Responses are scaled from “not free and fair”, “free and fair with major problems,” “free and fair 22 To conduct the analysis, I merged several datasets that include information on individuallevel and election-level measures. See the Appendix for an explanation of the steps involved in merging the datasets. 23 There are various advantages with using multilevel modeling (MLM) to analyze data that measured at different levels of analysis. For a comprehensive discussion of these advantages see Raudenbush and Bryk (2002). For the purposes of this analysis, MLM is utilized because it accounts for the potential problem of clustering in which a citizen’s election quality perceptions might be more similar to others in the same election period than with citizens’ views across different election periods. Additionally, MLM provides a more efficient estimation of the standard errors of the election-level predictors, relative to standard OLS regression. Be- cause of this MLM reduces the possibility of committing a type 1 error: that is, election-level variables are found to be significant when they actually are not. 24 More specifically I estimate a random-intercept logit model in which I assume that election-period intercepts vary based on my election-level predictors, including EMB performance and EMB design. The models are estimated using the xtlogit program in Stata 10. 130 with minor problems,” and “completely free and fair.” 25 I re-scale the original measure into a dichotomous indicator with 1 indicating that citizens perceived the elections as “completely free and fair,” or “free and fair with minor problems,” while 0 denotes the elections as “free or fair with major problems,” or “not free and fair.” 26 Across 28 election periods in the sample, a plurality of citizens (45 percent) rated the elections as completely free and fair, while a further 30 percent of those interviewed believed that the elections only had minor problems. When these two categories are combined approximately three in four Africans (75 percent) had a positive rating of the elections. Conversely, only 12 percent of those surveyed regarded the most recent elections as “not free and fair”, while 12 percent of respondents saw the elections as “free and fair with major problems” (26 percent). 25 In Round 1, there were two questions on election quality that differed by country: 1)“In your opinion, were the following elections conducted honestly or dishonestly?” Answers ranged from “Dishonestly” to “Very honestly;” and, 2) “On the whole, how would you rate the freeness and fairness of the last national election?” Answers ranged from “Completely free and fair” to “Not free and fair.” The first question was asked in five countries Ghana, Mali, Tanzania, Nigeria and Uganda while the second was asked in the remaining seven countries. While the wording for the questions and response options of the two election quality questions are different, they are considered functionally equivalent, especially because they have the same response scale. See Bratton et al. (2005) for the use of the election quality measure based on different questions across a 12-country sample in Round 1. To account for the potential bias that could emerge from the utilizing different question wording, in the regression analysis, I compare the results for models that exclude the countries with the second question type and the substantive results do not change (results not shown). 26 Although re-scaling perceived election quality as a binary variable may limit the variability in citizens’ evaluations, I employ the use of the measure because it provides better interpretation of the results. All empirical models using perceptions of election quality as the DV have been estimated using both the dichotomous and ordinal measures and there are no substantial differences in the results. Results of the ordered logit models using the STATA Xtlogit program are not shown, but are available upon request. Another potential concern is the reliability of using a single item indicator as the dependent variable; however, as pointed out by Anderson et al. (2005), the main problem associated with using a single item is lower significance levels. However, this will not bias the regression estimates. 131 Figure 4.3 displays the mean election quality perceptions across the 28 election periods in the sample. 27 At one end of the spectrum, citizens in Zambia’s 2003 election report the lowest levels of confidence (mean = 0.29), while citizens in Botswana’s 2004 election reported having the highest levels of confidence (mean = 0.84). Figure 4.3 also reveals that in some countries, including Ghana (1996 and 2004), Nigeria (1999 and 2003) and Zambia (1996 and 2001), mean election quality evaluations changed significantly across election periods. In other countries, including Botswana and South Africa, mean election quality perceptions remained relatively more consistent over time. 4.4.2 Independent Variables: EMB Design Of principal concern in this chapter is whether citizens’ evaluations of election quality (dependent variable) are influenced by EMB design—de jure autonomy and capacity— (independent variable) in Africa. EMB design reflects the formal rules and procedures as stated in law that shape the autonomy and capacity of EMBs. Similar to the previous section, I construct a composite index of EMB de jure autonomy that consists of six indicators: 1) institutional autonomy; 2) legal autonomy; 3) partisan autonomy; 4) commissioner appointment process; 5) commissioner tenure; and 6) financial autonomy. I also include a composite indicator of EMB de jure capacity based on five indicators: 1) professional capacity; 2) administrative capacity; 3) voter registration capacity; 4) conflict resolution capacity; and, 5) capacity to sanction political actors. A cursory examination of the data can provide some initial insights about the relationship between citizen’s election quality attitudes (dependent variable) and EMB de jure autonomy and 27 The mean election quality perceptions across the 28 election periods is 0.64 and standard deviation of 0.43 132 capacity indexes. Figure 4.4 displays average scores the de jure autonomy and de jure capacity indexes across elections that are either “free and fair” or “not free and fair.” First, the level of de jure autonomy seems to be the same in both types of elections (free and fair: 0.67; not free and fair: 0.66). However, the de jure capacity of EMBs seem to be marginally lower in free and fair elections (0.69 compared to 0.64). Already we see the two measures of EMB design—EMB de jure autonomy and capacity—may have little leverage in explaining perceived levels of election quality. 4.4.3 Election-level Controls Similar to the previous analysis, I also control for proportionality of the electoral system and the level of electoral violence. Along with these two main controls that are incorporated into the main empirical models, I also consider other factors that may influence citizens election quality perceptions in additional model specifications. These include the level of democracy, logged GNP per capita, time since the last elections, corruption perceptions index, and electoral fraud. 4.4.4 Individual-level Controls I also include a series of individual-level attributes that the literature highlights as being correlated with perceptions of election quality. The first, and most theorized, is citizens’ partisan affiliation. I operationalize this measure through a question from the Afrobarometer that asks citizens to identify the party to which they feel closest. I then construct three separate dichotomous variables: 1) Winning partisans—citizens who reported an affiliation with the winning party; 2) Losing partisans—citizens who indicated an affiliation with losing parties; and, 3) Non-partisans—citizens who reported no partisan affiliation. Following this, I incorporate an indicator of citizens’ level of political participation based whether the respondent reports having 133 voted in the last elections (Voted in Last Election). Most cross-national studies on election quality perceptions find that voters usually espouse more favorable election quality perceptions. As highlighted in the theoretical model, the level of political knowledge can mediate the formation of election quality perceptions. I account for citizens’ level of political knowledge through two separate measures. The first examines citizens’ frequency of obtaining news through the radio, TV, and newspaper sources. I construct an additive index of based on a factor analysis of the three measures of media exposure that is then normalized on a scale from 0 to 1, with 1 indicating higher levels of media exposure. I also include a measure of respondents’ level of education that ranges from no education (0) to post-graduate education (9). Following recent research in Africa, politically knowledgeable citizens are more likely to have greater insight on what free and fair elections should be, and more likely to recognize when elections deviate from these norms. In addition to media exposure and levels of education, I incorporate a composite measure for citizens’ level of political interest, based on two items which mea- sure citizens’ interest in public affairs and citizens’ frequency of discussing politics. I expect politically interested respondents to espouse more favorable election quality opinions because they are more likely to be affiliated with the incumbent political party (Moehler 2009). Lastly, I control for various social and demographic factors including age, gender and urban/rural residency 4.4.5 Results I consider the effect of EMB design through our measures of EMB de jure autonomy and capacity in Table 4.6. Models 1 and 2 display the results of the EMB de jure capacity and autonomy. In both models the coefficients are negative and statistically insignificant, providing some indication that neither dimension of EMB design can explain cross-national variations in citizens’ election quality perceptions, after controlling for other important covariates. Based on 134 this we fail to provide substantiation for either Hypotheses 2A or 2B. The nonsignificant findings for the effect of design on citizens, is similar to those found in the previous analyses of the effects on elites’ election quality perspectives. Moving to our control variables, the results indicate that proportionality of the electoral system has a positive but insignificant coefficient, while citizens in countries that experience more election violence had lower evaluations of election quality. Regarding the individual- level controls, opposition partisans and non-partisans (winning partisans: reference category), educated citizens, and those living in urban areas are less likely to consider elections free and fair. By contrast, citizens who reported having voted in the last election and citizens with high levels of political interests are more likely to believe that the last elections were democratic. 28 4.5 Discussion The results of the empirical analyses on the effect of EMB design on 1) elites and 2) citizens, indicate that neither EMB de jure autonomy or capacity are significantly associated with elites’ or citizens’ election quality attitudes. We can conclude from these findings that variations in EMB design cannot adequately explain perceptions of election quality across the countries included in the sample. The nonsignificant effect of EMB design, in the context of developing democracies, is not particularly surprising for a number of reasons. First, the model on the consequences of EMB design and performance (see Figure 1.2) provides some support for EMB design having little to no effect. One possibility is that institutional rules are weakly enforced and as a result, EMB rules do not influence the performance of EMBs, nor do they provide credible mechanisms to constrain the behavior of political elites or citizens. Weak enforcement 28 The coefficients for age and gender were not displayed in the table because they were not significant in the model. 135 and implementation of rules may be attributed to insufficient resources, lack of time to implement rules after reforms are enacted, or deliberate attempts by actors within and outside the EMB to block effective rule implementation. Another possibility is the existence other informal rules and procedures of election administration that either undermine (“competitive”) or enhance (“substitutive”) existing formal rules (Helmke and Levitsky 2006). The literature has examined various informal rules that either directly or indirectly shape EMB design. For instance, systematic clientelism and corruption have often nullified many of the rules guiding EMB autonomy in many African countries. In countries such as Nigeria, incumbent executives (president or governors) have been found officially culpable for using state resources to bribe electoral officials in return for rigging elections in their favor. Unfortunately, the indicators of EMB de jure autonomy and capacity do not effectively incorporate the effect of informal rules. A third reason highlighted in the theoretical model is that the political institutional context may undermine the effect of EMB design. Successful election administration depends on a network of impartial and well-resourced political and state institutions, including security agencies, judiciary, and state bureaucracy. In countries where state capacity is weak and there are not enough institutions to check executive power and electoral abuses, it becomes increasingly difficult for EMB design to have an impact on election quality. In addition to the theoretical model, the existing literature also provides some support for the absence of a significant link between EMB design and perspectives on electoral quality. Scholars of political institutions, especially those focusing on institutions within developing contexts, often find that the link between formal institutional rules and political behavior is difficult to capture empirically because the institutionalization of formal rules is an ongoing 136 process in some countries, while in others it has not yet started (Posner and Young 2007). For instance, Keefer and Stasavage (2003) find that central bank independence matters only in countries where the political and institutional context provides credible checks on the power of the incumbent. Other studies on political institutions in developing democracies such as legislatures (Barkan 2009) and courts (Eisenstadt 2004) have come to similar conclusions. 4.6 Conclusion Does EMB design (de jure autonomy and capacity) matter for political elites and citizens in Africa? Based on a sample of 19 African countries, the results of the regression analyses indicate that EMB design is not an important predictor of election quality perceptions. I argue that focusing only on EMB design tells only one part of the story because in many countries formal EMB rules are weakly enforced, undermined by dominant informal rules, such as clientelism or corruption, or they are nullified within institutional contexts characterized by an absence of checks and balances on executive control as well as low state capacity. This conclusion does not mean that institutions do not matter. It shows merely that formal institutional design, at least for EMBs, does not matter. It says nothing about the actual performance. This dimension may still matter. In the following three chapters (5, 6 and 7) I evaluate the extent to which EMB performance (de facto autonomy and capacity) may explain variations in election quality from the perspectives of political elites and citizens. 137 APPENDIX 138 Table 4.1: Factor Analysis of EMB De Jure Autonomy Indicators Factor analysis/correlation Method: principal factors Rotation: (unrotated) Factor Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Factor 6 Eigenvalue 3.70591 0.97695 0.62369 0.41165 0.17092 0.11087 Variable Institutional Autonomy Legal Autonomy EMB Appointment Process Tenure Length Financial Autonomy Partisan Autonomy Number of observations = 98 Retained factors = 2 Number of parameters = 6 Difference 2.72897 0.35325 0.21205 0.24073 0.06005 . Proportion 0.6177 0.1628 0.1039 0.0686 0.0285 0.0185 Factor 1 0.9065 0.8709 0.6554 0.7528 0.7496 0.7533 Uniqueness 0.1783 0.2415 0.5704 0.4332 0.4381 0.4326 139 Cumulative 0.6177 0.7805 0.8844 0.9530 0.9815 1.0000 Table 4.2: Factor Analysis of EMB De Jure Capacity Indicators Factor analysis/correlation Method: principal factors Rotation: (unrotated) Factor Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Eigenvalue 2.05441 0.22474 -0.02358 -0.03738 -0.21757 Variable Professional Capacity Administrative Capacity Voter Registration Capacity Conflict Resolution Capacity Sanctioning Capacity Number of observations = 95 Retained factors = 2 Number of parameters = 9 Difference 1.82966 0.24833 0.01380 0.18019 . Proportion 1.0269 0.1123 -0.0118 -0.0187 -0.1088 Factor 1 0.5280 0.8899 0.3995 0.5762 0.7015 Uniqueness 0.6093 0.2069 0.8282 0.6606 0.4159 140 Cumulative 1.0269 1.1392 1.1274 1.1088 1.0000 Table 4.3: Determinants of Election Quality (19-country sample) Model 1 Model 2 Model 3 Overall Election Overall Election Overall Election Variable Quality Quality Quality Ordered Logit Ordered Logit Ordered Logit EMB De Jure Autonomy -1.228 -1.223 — Index (1.642) (1.267) — EMB De Jure Capacity -0.02 — -0.782 Index (1.534) — (1.136) Proportional Electoral 0.155 0.17 0.116 Rules (0.498) (0.437) (0.477) Parliamentary Elections 4.824*** 4.802*** 5.010*** (1.558) (1.496) (1.527) Election Fraud -5.663*** -5.712*** -5.615*** (1.762) (1.709) (1.765) Election Violence -1.224 -1.301* -1.274 (0.789) (0.755) (0.781) Competitiveness of -0.0104 -0.00939 -0.0111 Elections (0.0122) (0.0120) (0.0122) International Observers -0.804 -0.778 -0.736 (0.766) (0.735) (0.760) Freedom House Index 1.819 1.871 1.804 (Lag) (1.724) (1.692) (1.731) Time since Last Election -0.0342 -0.0302 -0.061 (0.194) (0.191) (0.190) Pseudo R-squared 0.4552 0.4581 0.4514 Log Likelihood -41.245 -41.48 -41.53 N 84 86 84 Note: The analysis is conducted using xtlogit in Stata 10. Cell entries indicate the logistic regression coefficient and the standard errors in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1 141 Table 4.4: Election Periods, Countries and Afrobarometer Rounds Country Benin Botswana Botswana Cape Verde Ghana Ghana Kenya Lesotho Lesotho Madagascar Malawi Malawi Mali Mali Mozambique Namibia Namibia Nigeria Nigeria S. Africa S. Africa Senegal Tanzania Uganda Uganda Zambia Zambia Zimbabwe Year of Election 2003 1999 2004 2001 1996 2004 2002 1998 2002 2002 1999 2004 1997 2002 2004 1994 2004 1999 2003 1999 2004 2001 2000 1996 2001 1996 2001 1996 Afrobarometer Round Round 3 Round 1 Round 3 Round 3 Round 1 Round 3 Round 3 Round 1 Round 3 Round 3 Round 1 Round 3 Round 1 Round 3 Round 4 Round 1 Round 3 Round 1 Round 3 Round 1 Round 3 Round 3 Round 1 Round 1 Round 3 Round 1 Round 3 Round 1 142 Table 4.5: Citizens’ Evaluations of Election Quality: 28 Election Periods Response Option Percentage Frequency Not free and fair 11 4,874 Free and fair, with major problems 12 5,214 Free and fair, but with minor problems 27 6,095 Completely free and fair 40 13,366 Don’t Know/ Refused 10 19,827 Total 100 49,376 Note: “On the whole, how would you rate the freeness and fairness of the last national election held in XXXX? Cell entries represent percentage of respondents. Source: Afrobarometer Rounds 1,3 & 4, with Afrobarometer combined weights. 143 Table 4.6: Determinants of Citizens’ Election Quality Evaluations Model 1 -0.091 (0.405) Election-level EMB Design (De Jure Autonomy) Model 2   -0.608 (0.411) -0.736*** (0.216) (0.318) EMB Design (De Jure Capacity) Election Violence -0.729*** (0.226) 0.197 Proportional Electoral System Individual-level Controls Opposition Partisans -1.325*** -1.372*** (0.035) (0.036) Non-partisans -0.769*** -0.795*** (0.032) (0.033) Voted in Last Elections 0.332*** 0.333*** (0.030) (0.031) Index of Political Interests 0.153*** 0.142*** (0.043) (0.044) Index of Media Exposure -0.031 -0.053 (0.057) (0.059) Level of Education -0.049*** -0.052*** (0.008) (0.008) Urban Resident -0.103*** -0.103*** (0.030) (0.030) Random Intercept -0.880*** -0.833*** (0.276) (0.271) Individual Observations 40,462 38,359 Election Periods 28 27 Note: DV Perceptions of Election Quality (Afrobarometer Round 1, 3 and 4). The analysis is conducted using xtlogit in Stata 10. Cell entries indicate the logistic regression coefficient and the standard errors in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1 144 Figure 4.1: Theoretical Model: Causes of EMB Design and Performance Informal Rules Elite Attitudes & Behavior Formal Rules EMB Performance (De Facto Autonomy & Capacity) EMB Design (De Jure Autonomy & Capacity) Citizen Attitudes Political & Institutional Context 145 Figure 4.2: Mean EMB De Jure Autonomy and Capacity by Overall Election Quality 0.90 0.78 0.80 0.70 0.50 0.57 0.56 0.60 0.48 0.46 0.38 0.40 0.30 0.42 0.23 0.20 0.10 0.00 Not Free and Fair Significant Limited Free and fair (7) (12) Irregularities (86) Irregulariites (108) Autonomy Index Capacity Index 146 Figure 4.3: Citizens’ Evaluations of Election Quality by Election Period Zambia_01 Zimbabwe_96 Nigeria_03 Malawi_04 Lesotho_98 Mali_97 C.Verde_01 Ghana_96 Malawi_99 Mali_02 Zambia_96 Mean Uganda_01 Mozambique_04 S. Africa_99 S. Africa_04 Benin_03 Nigeria_99 Ghana_04 Namibia_94 Madagascar_02 Senegal_01 Lesotho_02 Namibia_04 Uganda_96 Kenya_02 Tanzania_00 Botswana_99 Botswana_04 0.20 0.30 0.40 0.50 0.60 0.70 0.80 Note: Mean Election Quality Perceptions: 1 = “completely free and fair” or“ free and fair with minor problems”; 0 = “free or fair with major problems” or “not free and fair.” Source: Afrobarometer Rounds 1,3 & 4, with Afrobarometer combined weights. 147 Figure 4.4: Variations in EMB Design by Election Quality 0.7 0.69 0.68 0.67 0.66 0.65 0.64 0.63 0.62 0.61 0.69 0.67 0.66 0.64 De Jure Autonomy Index Not Free and Fair De Jure Capacity Index Free and Fair 148 Variables and Coding EMB Design Coding Institutional autonomy (Original Coding): 0=Government Model; 1=Mixed Model; 2=Independent Model Legal autonomy (Original Coding): 0= not autonomous; 1= autonomy legally defined; 2= autonomy constitutionally defined. Partisan autonomy (Original Coding): Whether the commission is composed of partisan members or experts. 0=Partisan membership, 1=mixed membership, 2= Expert membership. Appointment process of EMB commissioners (Original Coding): Appointment process for EMB commissioners as stated in law: 1= Legislature only or executive only 2= Mixed appointment: legislature and executive (legislature w/ more than 50%– opposition support) 3= Mixed appointment: State entity other than legislature and legislature/executive 4= Mixed appointment: politicians and civil society/ judiciary 5= Judiciary: Supreme Court or Council of Judges 6= Civil society: Commission of lawyers and academics, etc. 7= Competitive Recruitment / Appointment by International Body Tenure length of EMB commissioners: (Original Coding): 0= No tenure; 1= 0-1 years of tenure; 2= 2-5 years of tenure; 2-5 years of tenure (possibility of renewal); 6-9 years of tenure; 6-9 years with (possibility of renewal); 10 or more years of tenure Financial autonomy (Original Coding): Commission Control over operational budget 0= not stated in law; 1= Budget submitted to a Ministry; 2=Mixed budget submissions; 3= Budget submit- ted directly to Parliament) Professional capacity (Original Coding): 0= No qualifications stated; 1= No legal or EMB expertise stated; 2=At leas one member of the commission should have legal training Administrative capacity (Original Coding): 0= Temporary: the secretariat is created before elections; 1=Mixed: more than half of the staff are seconded from other civil society ministries; 2=Permanent Voter registration capacity (Original Coding): 0=No register; 1=paper-based register; 2= computerized register (OMR); 3= electronic/ biometric data capture 149 Conflict resolution and political-party cooperation capacity (Original Coding): Existence of a system of communicating with Political Parties during elections. 0=No PPA; 1= PPA created through convention; 2= PPA created through law Sanctioning capacity (Original Coding): Main EMB has the power to sanction electoral offenders 0= No power to Sanction; 1= complete power to sanction EMB de jure autonomy index (Original Coding): 0-1 with increasing levels of de jure autonomy. I first normalize the six indicators on a 0-1 scale with 1 representing greater autonomy. Next, I conduct a factor analysis (principal components with oblique rotation) using the six EMB autonomy indicators. EMB de jure capacity index (Original Coding): 0-1 with increasing levels of de jure capacity. I first normalize the five indicators on a 0-1 scale with 1 representing greater autonomy. Next, I conduct a factor analysis (principal components with oblique rotation) using the five EMB capacity indicators. EMB Performance EMB Election Performance Index (Quality of Elections: Kelley 2011): weighted index of two variables that measure various de facto autonomy and capacity of the EMB during the pre-election period and on election day, the index has been normalized on a 0-1 scale with 1 indicating high performance. The pre-election performance indicator measures problems with voter registration process, appointment of election commissioners; enforcement of election law by commission; level of commission’s trans- parency; voter/candidate information and education; technical /procedural problems with ballot design and printing. The election performance measures logistical problems at voting locations – large crowds, long wait, confusion, inadequate information on voting procedures, inadequate facilities for voting. Human resource deficiencies – voting locations opening late, poll officials’ unfamiliarity with election laws or inconsistent application of voting procedures, and complaints about the conduct of election officials. Voter–list problems – missing names, fictitious and names of deceased on list, list not at polling station on time. 150 Aggregate-level Measures Election Fraud (Quality of Elections). Additive index of two variables that measure various dimensions of EMB performance during the pre-election period and on election day, the index has been normalized on a 0-1 scale with 1 indicating high fraud. Quality of Legal Election Framework (Quality of Elections). Measures the quality of the legal frame- work for holding elections. 0= High quality; 5=Low quality. International observers (Quality of Elections): 1= if the report mentions the assessment of inter- national observers; 0= otherwise. Overall Election Quality (Lindberg): It captures whether electoral procedures were followed, whether people could join a political organization of their choice, whether demonstrations were allowed, and information freely available. We have a four-point scale for this indicator: 0 – “No” when no elections have been held or those held have been wholly unfair; 1 – “Irregular” when there were serious defects that influenced the results; 2 – “Yes, irregularities not significant” when there were deficiencies but they did not affect the result; and 3 – “Yes” when elections were free and fair. Opposition participation (Lindberg): Did the real opposition participate in the elections: 0= no; 1=some but not all of the main players; 2=yes Losers Accept Results (Lindberg): Did the losers accept the results? 0=no; none of the main players; 1=not at first but later OR some but not all; 2=yes; all main players immediately Election Violence (Lindberg): Did election campaign/day violence occur? 0= High quality; 5=Low quality. 2=yes to a significant extent; 1=yes, but only isolated incidents; 0=no, peaceful. Turnover in Power (Lindberg): Did the elections result in a turnover of power? 0=no; 1=half: new party or person; 2=yes Type of election (Lindberg): 1= Parliamentary system; 0=Presidential Proportional Representation (Lindberg). Parliamentary Electoral system 0= Plur/Maj; 1=PR/Mixed Freedom House Index (Quality of Governance). Weighted index of freedom house civil liberties and political rights sub-scores. 0-1 increasing democracy. Log GDP PPP (World Bank Indicators). Log of gross domestic product per capita. 151 Individual-level Measures Free and Fair Elections (Afrobarometer) : On the whole, how would you rate the freeness and fairness of the last national election. 0= Not free and fair and Free and fair, with major problems; 1=Free and fair, but with minor problems and Completely free and fair; Missing = Do not understand question, Don’t know, Refused. Partisan Affiliation (Afrobarometer) : Do you feel close to any political party. Used three dummy variables to represent the three categories. Winner : 1= Affiliation with winning party; 0= Losers and Non-partisans. Losers : 1= Affiliation with losing party; 0= Winners and Non-partisans. Non-Partisans : 1= Affiliation with no political party; 0= Winners and Losers. Voted in last elections (Afrobarometer) : With regard to the most recent, [last] national elections, which statement is true for you? 1=Voted in elections; 0= Did not vote in elections. Political Interest Index (Afrobarometer) :How interested would you say you are in public affairs? When you get together with your friends or family, would you say you discuss political matters. Two item composite indicator normalized on a 0-1 scale: 0=low interest; 1= high interest. Index of Media Exposure (Afrobarometer) : How often do you get news from the following sources: Radio, TV, Newspaper (Three item additive index 0=low exposure; 4=high exposure) Education (Afrobarometer): Level of education of respondent (0=no education 9=post graduate). Social Structure: Gender: 0=Male, 1= Female. Location of Respondent: 0=Rural, 1=Urban. IFES Post Election Survey Free and Fair Elections: In your opinion, how free and fair were the 2007 Presidential and Parliamentary elections? Were the 2007 Presidential and Parliamentary elections completely free and fair, somewhat free and fair, not too free and fair or not at all free and fair? 1= completely free and fair/Somewhat free and fair, 0= Not too free and fair /Not free and fair; Missing = Do not understand question, Don’t know, Refused. 152 INEC Performance [Police, Military, EFCC, Courts]: In your opinion, how effective are each of these institutions and leaders in carrying out the duties that are their responsibility? Are these institutions and leaders very effective, somewhat effective, not too effective, or not at all effective. 0 = Not at all effective, 1= Not too effective, 2= Somewhat effective, 4= Very Effective. INEC Autonomy Index: Please tell me how satisfied or dissatisfied you were with the following aspects of the 2007 Presidential and Parliamentary elections [Impartiality of polling station staff, Impartiality in the counting, tabulation and announcement of results]. First of all, thinking about the performance of the Independent National Electoral Commission (INEC), with which of the following statements do you agree more: The Independent National Electoral Commission (INEC) performs its duty as a neutral body guided in its work only by the law, OR INEC makes decisions which favor particular people or interests? Weighted average of the 3 items rescaled from 0-1: 0= low autonomy; 1=high autonomy. (Chronbach’s alpha: 0.752) INEC Capacity Index: Please tell me how satisfied or dissatisfied you were with the following aspects of the 2007 Presidential and Parliamentary elections [Voter Registration Process, Information on voting procedures and how to mark the ballots, Competency of the polling station staff and Adequacy of facilities and equipment at the polling station]. Weighted average of the 4 items rescaled from 0-1: 0= low capacity; 1=high capacity (Chronbach’s alpha: 0.829). Partisanship : Can you tell me which political party best represents aspirations of people like you? Used three dummy variables to represent the three categories. Winner : 1= Affiliation with winning party; 0= Losers and Non-partisans. Losers : 1= Affiliation with losing party; 0= Winners and Non-partisans. Non-Partisans : 1= Affiliation with no political party; 0= Winners and Losers. Election Intimidation Experience: On or before the presidential and parliamentary elections, did anyone try to pressure YOU to vote for a certain candidate in the election? 1= Yes; 0=No Vote buying Experience: On or before the presidential and parliamentary elections, did anyone try to pressure YOU to vote for a certain candidate in the election? 1= Yes; 0=No. Election Violence Experience: Did you see or hear of any violence related to the Presidential or Parliamentary elections. 2= Saw violence; 1= Heard violence; 0=No experience w/ violence. 153 Voted in last election: Did you vote in the April Presidential Elections or Not. 1=voted in presidential elections; 0= Did not vote Political Interest: Please tell me how interested you are in matters of politics and government. 0= Not at all interested; 1=Not too interested; 2= Somewhat interested; 3=Very interested. Media Exposure: Please tell me how often you [Watch television, read newspapers, listen to the radio] for news on politics and government— every day, a few times a week, a few times a month, once a month or less OR never? 0= Low media exposure; 1= High media exposure (Chronbach’s alpha: 0.713). Knowledge of Election Processes: In general, how much information would you say you have about how the recent Presidential and Parliamentary elections were organized and run? Do you have a great deal of information, a fair amount, not too much or none at all? 1= Great deal/Fair amount; 0= Not too much/None at all. Social Structure: Education: 0=No formal schooling, 9=Post-graduate. Gender: 0=Male, 1= Female. Location of Respondent: 0=Rural, 1=Urban. Performance of [International Observers, Domestic Observers, Local INEC]: For each of the following, please tell me how effective or ineffective you think this was in helping to ensure the Presidential and Parliamentary elections were free and fair. Was it very effective, somewhat effective, not too effective or not at all effective? 0 = Not at all effective, 1= Not too effective, 2= Somewhat effective, 4= Very Effective. 154 CHAPTER 5 THE CONSEQUENCES OF EMB PERFORMANCE ON ELITES’ ELECTORAL ATTITUDES AND BEHAVIOR 5.1 Introduction The Opposition refused to accept the results of the presidential election because we believe and still do that the presidential election was (and not must have been) rigged ... there was a deliberate, pre-arranged plan by the PNDC in collusion with INEC, to rig the 1992 November presidential election in favor of PNDC/NDC candidate, Jerry John Rawlings. (Adu Boahen, National Patriotic Party Presidential Candidate, 1992 Ghanaian Presidential Elections) In Chapter 3 we learned that, in many African countries, there is an ongoing struggle among political elites over the rules of electoral administration. Domestic and international actors consider electoral management bodies an important battleground because of the strategic role of the institution in shaping the quality of elections and pace of democratization. The introductory quotation presents the perspective of Professor Adu Boahen, the National Patriotic Party (NPP) presidential candidate in the 1992 Ghanaian elections. The NPP failed to win the struggle over the rules shaping the electoral administrative body in Ghana (Interim National Electoral Commission) during the run-up to the elections. Professor Adu Boahen cited INEC’s lack of political autonomy and the absence of a credible voter register as the main reasons for challenging the results of the Presidential elections and the NPP’s boycott of the Parliamentary elections scheduled for later that year. The chapter proposes a theory, based on historical institutionalism, to explain the relationship between EMBs and elites’ electoral attitudes and behavior. Following the arguments 155 of Professor Boahen and the NPP, I suggest that EMBs matter for elites, in emerging democracies, because the performance of the institution directly influences the likelihood of 1 electoral manipulation and the incidence of electoral administrative irregularities. Following the previous chapter, I distinguish between two dimensions of EMB performance: autonomy and capacity. I argue that variations in EMB autonomy and capacity shape the strategic interaction of elites as well as their evaluations of electoral integrity. When EMBs enjoy high levels of autonomy and capacity, the incumbent—and political elites more generally—find it more difficult to manipulate the EMB and engage in more general acts of electoral fraud. Moreover, EMB officials are better able to resolve the challenges of election administration and in so doing, reduce the incidence of electoral administrative irregularities. All things being equal, I expect de facto EMB autonomy and capacity to have positive implications for the strategic interaction between elites and the overall quality of elections. Mainly, political elites are more likely to play by the rules of the game (and eschew fraud and violence), participate in elections, and accept electoral defeat without crying “Foul!” I examine the applicability of this argument using two empirical approaches. First, I incorporate perspectives of elites in Ghana, Nigeria, South Africa, and Zimbabwe to understand the various ways in which EMBs may shape the elite behavior and attitudes toward election quality during recent elections. The analysis is based on original data collected through in-depth interviews between 2010 and 2012 with 88 elites (key stakeholders involved in the election process) in the four countries. Elites, for the most part, regarded the performance of EMBs as the most important determinant of their election quality evaluations and electoral behavior. 1 I distinguish between electoral manipulation and electoral administrative irregularities. Electoral administrative irregularities are conceptualized as any systematic deficiency in election administration that is not a consequence of intentional manipulation. 156 Furthermore, elites frequently distinguished between EMB autonomy and capacity. However, the attention they placed on these dimensions differed across countries. For instance, autonomy seemed to take on greater levels of importance in countries such as Nigeria and Zimbabwe in which previous elections had fallen short of democratic principles. Additionally, elites identified the gap between the formal rules that influenced the design of EMBs and the actual EMB performance. Although they recognize the importance of EMB design, most elites explained that actual EMB performance during elections had the greatest bearing on their actions. Finally, while elites viewed independent and competent election management as necessary for electoral integrity, most agreed that EMBs were not sufficient to guarantee the democratic quality of elections. Instead, they highlighted the importance of an institutional context characterized by a network of professional and politically neutral state institutions (judiciary, courts, security agencies, and bureaucracy) to support the operations of EMBs. I follow this rich qualitative analysis with a more systematic assessment of the link between EMBs and election quality using cross-national data. I first identify four proxies of election quality: elite perceptions of election quality, incidence of electoral boycotts, losers’ acceptance of results, and electoral turnover. Next, I define EMB performance as the actual autonomy and capacity that the EMB displays during election (de facto EMB autonomy and capacity). Then, I evaluate the relationship between the four indicators of election quality (dependent variables) and EMB performance (independent variable) across 45 African countries between 1990 and 2004. The results indicate that high levels of EMB performance (de facto autonomy and capacity) are positively associated with elites’ perceptions of free and fair elections, electoral turnover, and losers’ willingness to accept defeat. 157 Generally speaking, these results have some important implications for how we understand the role of EMB performance and design in Africa. First, the robust relationship between EMB performance (de facto autonomy and capacity) and various election quality indicators in the statistical analyses demonstrates that EMBs matter. When these results are combined with the findings from the elite interviews, we can confidently assert that EMBs play an important role in shaping the quality of elections and the democratization process in Africa. This has important consequences for the policy community. Specifically, it highlights the challenges associated with promoting EMB reforms and the need for greater focus on establishing effective electoral management through a network of democratic institutions. The following section of the chapter outlines the main theory linking EMBs to elite behavior. Sections 5.3 and 5.4 provide an empirical examination of the theory while Section 5.5 concludes. 5.2 Consequences of EMBs for Elites: Theoretical Approach 5.2.1 Historical Institutionalism I rely on historical institutionalism to explain the relationship between electoral management bodies and elite behavior (Thelen and Mahoney 2009). Proponents of historical institutionalism assume that institutions reflect a combination of formal and informal rules which 2 provide incentives that shape the behavior of rational political actors (North 1990). Institutions 2 For the purpose of this analysis, I focus on three main elite actors: incumbent, opposition parties, and EMB officials. All three actors focus on two important features of the election environment: 1) the incidence of electoral manipulation and, 2) the extent of electoral administrative irregularities. Incumbent and opposition parties are office-seeking and respond strategically to the design and performance of EMBs and the expected behavior of other actors. EMB officials instead are motivated to reduce the electoral irregularities and ensure that political elites play by the rules of the game. 158 such as EMBs provide vital information on the expected behavior of other actors as well as the effectiveness of enforcement mechanisms to ensure that actors play by the rules. Another important feature of institutions is that they provide coordinating mechanisms that help reduce the transaction costs to cooperation. These informational and coordinating features are particularly important for our assessment of EMBs in emerging democratic contexts. In regimes where there is a legacy of overbearing authoritarian rule and the current institutional context fails to provide sufficient checks against executive manipulation, EMBs can serve as institutions that will guarantee the “procedural legitimacy” and the “substantive uncertainty” of the elections (Mozaffar and Schedler, 2002:11). In other words, EMBs help provide institutionalized democratic uncertainty; that is, the outcomes of elections are not pre-determined. 5.2.2 Characteristics of EMBs: Autonomy and Capacity There are various dimensions of EMBs that elites may consider important when evaluating these institutions’ ability to reduce electoral fraud and minimize electoral irregularities. I choose to focus the overarching and interrelated dimensions of 1) autonomy, and 2) capacity. At the theoretical level autonomy and capacity are both dimensions of the institutionalization of EMBs. One reason for distinguishing between EMB autonomy and capacity is that existing scholars have made a similar conceptual distinction in research on the 3 4 state , bureaucracies, and other political institutions including the legislature and courts. A second reason is more empirical: as highlighted in Chapter 3, there are differences in the level of EMB autonomy and capacity within specific countries. Third, the distinction is appropriate 3 4 See Bratton 1994; Geddes 1994; Skocpol 1985. See Hammond and Knott, 1996; Huber and McCarty 2004. 159 because variations in EMB autonomy may have different consequences for elite assessments of the electoral environment relative to variations in capacity. As will be highlighted in the empirical section below (Section 3), elites across four African countries place a different emphasis on how these characteristics may affect the quality of elections. For instance, elites in Nigeria believed that replacing the then-existing electoral commissioner, Professor Maurice Iwu, with a more impartial and credible commissioner was a necessary condition for free and fair elections in 2011. Meanwhile, Ghanaian elites expressed the importance of developing an advanced voter registration system as essential for the credibility of 2012 elections. What is most important is that both understandings of EMB autonomy and capacity are instrumental to gauging elite attitudes and behavior. Pastor highlights this point clearly: An [EMB] that is independent, impartial, authoritative, and competent and perceived as such, with adequate resources, has a far greater likelihood of conducting an election that is judged fair and free by all parties in a country and by the international community than one that does not have these attributes (1999: 17). 5.2.3 EMB Autonomy EMB autonomy is an important characteristic because it provides some indication of the extent to which the incumbent, as well as other political elites, can manipulate the EMB and 5 engage in other forms of electoral fraud. In the words of the two leading scholars on electoral management “establishing an independent electoral commission, in fact, has become a compelling international norm, a sine qua non of electoral credibility” (Mozaffar and Schedler 5 In many emerging democracies with a legacy of strong executive powers, the executive represents the greatest threat to the autonomy of political institution, including EMBs. As a result I assume that electoral administrative manipulation is most likely to be perpetuated by the incumbent and members of his political party. 160 2002: 15). The overwhelming need for EMB autonomy within many African countries can be linked to the legacy of powerful executives who often utilize a “menu of electoral manipulation” to ensure electoral survival. Another feature of African elections is the do-or-die, zero-sum political culture in which political elites pursue electoral victory at all cost (van de Walle 2002; Schedler 2002a). In these political contexts, EMB officials are constantly exposed to monetary inducements, promises of political appointments, and threats of sanctions to rig elections in favor of a particular party or candidate. As a result there is always a palpable threat that EMBs will be politically manipulated. First, I suggest that greater levels of EMB autonomy raise the costs for the executive to manipulate EMB operations and decision-making. Recent studies of EMB autonomy have incorporated traditional principal-agent theory to show how delegation of electoral management from the executive (the principal) to independent bodies (the agent) reduces the mechanisms through which the executive can influence the administration of a process in which he/she has a vested interest (Magaloni 2006b; Rosas 2010; Birch 2011). The most straightforward ex- ample of this is seen with the different models of electoral management. As you may recall, EMBs can either be government-run, independent, or a combination of the two. According to Lopez-Pintor (2000), electoral management bodies are less likely to be influenced by the executive when they are institutionally independent of the executive branch of government. Because the majority of EMBs in Africa are nominally independent, executives can influence the preferences of EMB officials primarily through the procedures that stipulate the appointment, tenure, removal, and remuneration of EMB officials. Trebicock and Chitalkar (2009) explain how crucial shortcomings in the institutional autonomy of the Electoral Commission of Kenya (ECK) and the Zimbabwean Electoral Commission (ZEC) allowed for the executive to significantly manipulate 161 both institutions during the 2007 and 2008 elections (respectively). 6 Conversely, Ghana’s Electoral Commission (EC) withstood concerted attempts of incumbent- led manipulation during the 2008 elections which resulted in the victory of the main opposition party (Debrah 2011). Leading Ghanaian scholars attribute the EC’s integrity and impartiality to certain constitutionally enshrined statutes, such as the life-long tenure of the chair/deputy chairpersons that underscore the institution’s autonomy (Debrah 2011; Gyimah-Boadi 2009). Second, incumbents may find it more costly to engage in other forms of electoral manipulation (such as systematic vote buying, violence and intimidation, and stuffing of ballot boxes) when electoral administrative rules endow EMBs with greater autonomy. In most countries, EMBs are mandated to promote free and fair elections by ensuring that political actors adhere to prevailing domestic electoral laws and international commitments. In this respect, EMBs serve as institutions of horizontal accountability by deterring and sanctioning electoral malpractice (Schedler, Diamond and Plattner 1999; Gyimah-Boadi 1999). 7 In effect, EMBs enhance the integrity of the democratic process by providing checks and balances against the executive and other political elites. However, the extent to which the EMB can effectively deter other forms of manipulation and exercise discretion depends on the extent of its autonomy. EMB officials have more incentives to hold the political elite accountable to electoral laws when they 6 The authors detail how in Kenya the ECK lacked institutional, financial and operational autonomy, while in Zimbabwe these factors plus the ZEC’s limited powers undermined its ability resist executive manipulation. 7 Hounkpe and Fall et al. note that Nigeria’s Independent National Electoral Commission (INEC) has the power to disqualify a candidate if that candidate engages in electoral irregularities. In Ghana the EC can invalidate election results before their announcement when extensive irregularities have occurred (2010: 68-72). 162 are comprised of members who do not necessarily share the preferences of the incumbent (or any political party) and when they enjoy some degree of legal protection from executive retaliation. One excellent example of the effect of autonomy can be seen through the tenure of EMB commissioners as stated in law. Constitutions and electoral codes often indicate the length of time for which EMBs commissioners are appointed. Tenure appointments range from life- long tenure appointments in Ghana to appointments for four years in Malawi and appointments for the election-period (6 months) in Niger. In a recent interview with, Dr. Afari-Gyan, the current Chairman of the Electoral Commission of Ghana he explained how life-long tenure has enhanced the autonomy of the EC in Ghana by insulating the commissioners from the influence of the incumbent and other dominant state and society actors. 8 For him, the security of tenure meant that he had the discretion to make decisions that were unfavorable to the ruling political party of the day without worrying about being arbitrarily removed from office. 9 Third, opposition elites also respond strategically to the perceived extent of EMB autonomy and expected behavior of incumbents. If opposition elites believe that electoral administrative rules provide credible mechanisms to deter the manipulation of the EMB (and other forms of electoral fraud), the EMB may alter the incentives and behaviors of the opposition elite and reduce the uncertainty in the electoral environment. One potential consequence of greater EMB autonomy is that opposition elites have greater incentives to play by the rules of the game and not engage in their own electoral fraud and manipulation. While the incentive for 8 Personal interview, Kumassi, Ghana, May 2010. 9 In many ways, we can compare autonomous EMBs to international election observer groups, which according to leading scholars, are effective in altering executives’ incentives to cheat (Kelley 2012; Hyde 2011). This is achieved by increasing the risks of exposing electoral fraud. 163 engaging in fraud can never be completely eliminated, Pastor (1999) argues that opposition elites have interests in ensuring that elections are free and fair. A second impact of EMB autonomy is that opposition elites will be more willing to participate in elections when EMBs have autonomy. Numerous scholars have linked the likelihood of electoral boycotts to the independence of EMBs and the degree of electoral administrative irregularities (Bratton 1998; Lindberg 2006; Beaulieu and Hyde 2009; Kelley 2011). 10 Cameroon illustrates this point clearly. Opposition parties decided to boycott the National Assembly elections when the government refused to alter the composition of the electoral commission (Pastor 1999). Similarly the government’s refusal to establish an independent commission in Mauritania led to opposition boycotts before the December 1997 presidential elections (Beaulieu and Hyde 2009). 11 Other instances of opposition boycotts linked to opposition doubts about EMB autonomy, include Ghana (1992), Nigeria (2003), Zimbabwe (2008). While opposition parties may not want to bear the legitimacy costs of not participating in elections, they can register protests by failing to accept the results of the elections (Lindberg 2006b). Therefore, a third way in which EMB autonomy might affect elite behavior is by enticing elites to accept the results of the elections. One way in which opposition parties 10 Decisions to boycott are also tied to more general issues of electoral fraud and manipulation of electoral rules that occur early within the election period or after one round of an election (Kelley 2011). There is always the possibility that parties boycott elections they believe they cannot win or boycott perfectly fair elections. However, most scholars believe that boycotts are primarily motivated by electoral irregularities: even in unfair elections some parties will decide against boycotting because the associated costs are too high (such as losing representation in parliament and losing popular legitimacy). 11 In Cameroon two opposition parties, the Social Democratic Front (SDF) and Cameroon Democratic Union (UDC), boycotted the poll. All the main opposition parties boycotted the Mauritanian elections. 164 officially challenge election results is by bringing cases before the courts or electoral tribunals. A case in point is Nigeria. After the 2007 elections, 1,200 petitions were filed with the Election Tribunals. For many of the cases, losing political candidates (parties) cited overt collusion between EMB officials and political candidates to rig the elections as the main justification for overturning the official results. 5.2.4 EMB Capacity Having an autonomous EMB is not sufficient for the electoral process to effectively reflect the preferences of the voters. In addition to achieving impartiality and political neutrality, an- other core function of the EMB is to guarantee that elections are efficiently organized. This is best realized when EMBs possess professional and highly trained permanent and temporary staff, adequate sources of funding, sufficient supply of polling materials, and the requisite technical and logistical capacity to count millions of votes. It goes without saying that organizing elections is a complex endeavor especially in emerging democracies. Because elections involve the largest peacetime mobilization of a country’s resources, minimizing electoral irregularities represents a difficult task, and it becomes even more challenging with the structural, financial, and humanresource challenges faced by many developing countries in Africa (Mozaffar and Schedler 2002; Elklit and Reynolds 2000). The attitudes of political elites toward the quality of elections are closely associated with the perceived capacity of EMBs to administer elections effectively. This is because shortcomings in EMB capacity often result in administrative irregularities that undermine the integrity of the elections. It is also possible that deficiencies in EMB capacity may also increase the opportunities for the executive to engage in electoral fraud. 165 How do elites respond to deficiencies in EMB capacity? First, their reactions depend on the type of capacity deficit and the extent to which it can be remedied without compromising the integrity of the elections. Additionally, elites are concerned with whether the capacity deficits are intentionally orchestrated or are instead the consequence of poor planning and unexpected logistical challenges. While in many countries, political elites have intentionally undermined the capacity of EMBs for their electoral benefit, Bratton explains that opposition actors have a tendency to assume the worst and believe that the problems are politically motivated, especially during the first or second phases of multi-party elections (1998). Overall, we should expect that challenges with EMB capacity should reduce elite confidence in the integrity of elections. If these problems are systematic enough to adversely affect the outcome of elections, elites may either boycott elections or use these issues as the basis for disputing the results of the elections. 5.2.5 Sources of EMB Autonomy and Capacity There are two ways in which elites can develop their assessments of EMB autonomy and capacity. One way is by examining EMB design (formal rules that stipulate the organization and function of EMBs). The other is by assessing EMB performance (the implementation of these rules and procedures in practice). As discussed in Chapter 4, one key aspect of EMB design involves the rules that establish the de jure autonomy or de jure capacity. The de jure autonomy or capacity of EMBs provide elites with an indication of how EMBs might be expected to perform during the elections. For instance, the appointment process for electoral commissions, as stated in law, provides one excellent example of an attribute of de jure autonomy that elites often use to gauge the overall autonomy of the institution and its ability to make decisions independent of executive influence. The greater the number of political actors involved in the appointment process, as stated by law, 166 the less elites believe that the EMB commissioners are hand-picked by the incumbent to do his/her bidding. One would expect elites in South Africa (during elections since 1999) would have greater confidence in the autonomy of the EMB when compared to Kenyans (during the 2002 and 2007 elections) because EMB commissioners in South Africa are selected by the president based on a list of nominees provided by multiple state and non-state actors, 12 while in Kenya appointments are made solely by the president. Importantly, however, elites are not only concerned about the appointment process as stated in law but also whether or not the process was followed in practice. As a result, we would expect elites to be responsive to EMB performance or the de facto autonomy and de facto capacity when compared to EMB design. First, from a theoretical perspective contemporary African scholars have noted that the institutionalization of formal rules is weak. They highlight the legacy of neopatrimonial rule associated with strong executive powers, pervasive corruption, and enduring patron-client networks as important causes for disregard for rule of law (Posner and Young 2007). As a result, in countries where rule of law is weak, the de jure EMB autonomy and capacity may not accurately capture the EMBs’ ability to deter executive manipulation and minimize administrative problems. In his research on electoral courts in Mexico, Eisenstadt (2002) highlights the disparity between formal rules and EMB performance. In his judgment, rules only became important for electoral courts when these courts achieved greater insulation from the government and secured the confidence of opposition parties. Gazibo’s pioneering study of EMBs in Africa highlights a 12 South Africa’s electoral commissioners are appointed by the president upon recommendation of a committee of the National Assembly. Recommendations are made from a list of candidates prepared by a panel of representatives from the other institutions supporting democracy, namely the Human Rights Commission, the Constitutional Court, the Commission on Gender Equality, and the Public protector. 167 similar gap between EMB autonomy based on formal-legal provisions and the performance of EMBs (or what he refers to as “empirical” autonomy). Gazibo explains that “empirical autonomy ... is the result of power politics between political actors engaged in the political struggles that arise during the process of democratization” (620). On one hand, there are EMBs that are formally autonomous but “empirically” powerless. On the other hand, there are those EMBs that started with low levels of de jure autonomy and de jure capacity but that have enhanced their actual autonomy and capacity over time. Moreover, cross-national studies on EMB autonomy and capacity have emphasized the importance of distinguishing between formal rules and EMB performance (see Rosas 2010; Birch 2008; Birch 2011). My own research has revealed several reasons for making the distinction between design and performance with respect to EMBs in Africa. First, incumbents and political actors intentionally undermine EMBs by failing to implement or adhere to legislation concerning electoral management. Even though on paper the electoral commission in Nigeria may have been financial autonomous in 2007 through the passage of an electoral amendment, the incumbent allegedly used his powers over the national coffers to constrain the actions of EMB officials. As a result, opposition parties and the public in general believed that INEC lacked sufficient funding to purchase necessary election materials in time for the election (EU EOM 2007). As Mozaffar and Schedler suggest, “formal institutions matter. But even if they look nice on pa- per they have to prove in practice that they matter before actors start taking them seriously” (2002:21). Second, in some countries, legal reforms pertaining to EMBs are made immediately before elections without sufficient time for implementation. For this reason, the rules may not adequately reflect the performance of the institution during the election because elites and citizens do not have adequate time to familiarize themselves with changes in the rules. 168 5.2.6 Observable Implications and Hypotheses The previous discussion highlights some observable implications of the relationship between EMB autonomy and capacity and elites’ electoral behavior. First, I expect that countries where institutionalized EMBs perform well (that is, they display greater autonomy and capacity) experience more free and fair elections because political elites have less incentive to engage in fraud and violence and because EMB officials are better able to minimize administrative problems. Therefore, • (H1) When EMBs perform well, political elites are more likely to consider elections free and fair Not only should we find that elections have greater integrity when administered by competent and autonomous EMBs but there might also be consequences for the likelihood of opposition participation, for losers’ acceptance of the election results, and for the likelihood that opposition parties win the elections. Hence, I propose three additional hypotheses: • (H2) When EMBs perform well, opposition elites more likely to participate in elections • (H3) When EMBs perform well, losing elites more likely to accept election results • (H4) When EMBs perform well, there is a greater likelihood of electoral turnover In the following sections (4 and 5), I examine these hypotheses using recent qualitative data on elite perspectives in Ghana, Nigeria, South Africa and Zimbabwe. I complement this with a cross-national analysis of elite attitudes and behavior across 45 African countries. 5.3 Consequences of EMBs for Elites in Ghana, Nigeria, South Africa and Zimbabwe 169 Between 2010 and 2012, I conducted 88 semi-structured, face-to-face interviews with political party executives, leaders of domestic and international NGOs, journalists, and EMB commissioners and directors in Ghana, Nigeria, South Africa and Zimbabwe to assess the consequences of EMB performance on election quality. Elites in all four countries regarded the autonomy and capacity of electoral commissions as both necessary but insufficient conditions for free and fair elections. Other important conditions included the impartiality and competence of other institutions of election management; these included security agencies, bureaucracy and the judiciary. While elites clearly distinguished between EMB autonomy and capacity, ensuring EMB autonomy was most important in countries that were beset by electoral fraud and political manipulation. Finally, elites were cognizant of the difference between design and performance. While electoral administrative rules are important, elites were most concerned with the performance of the EMB when evaluating the quality of elections. 5.3.1 Role of EMBs in Elections When asked to explain the factors that influence the quality of previous elections, an over- whelming majority of elites highlighted the important role of EMBs. Most elites when interviewed identified the effect of EMB performance on their confidence in the voting process. 13 For instance, Ghanaian election stakeholders praised the Electoral Commission (EC) for promoting confidence among opposition and incumbent elites during the 2008 presidential elections; this was one of the most competitive elections in the country’s history and it resulted 13 Details about the types of elites interviewed for each country, the sampling procedures, and the representativeness of the sample are outlined in Chapter 3. 170 in a turnover in political leadership. 14 In contrast, the performance of Nigeria’s Independent National Election Commission (INEC) during the 2007 elections was so abysmal that most stakeholders, even senior members of the ruling party noted that they had lost confidence in the institutions’ ability to conduct elections. Furthermore, I found that EMB performance was also instrumental in understanding whether political elites participated in elections or eventually accepted the final election results. For in- stance, some Zimbabwean opposition party officials and civil society leaders blamed the inability of the Zimbabwean Electoral Commission (ZEC) to announce the results of the firstround in the presidential election (2008) as one cause for the political intimidation experienced by opposition supporters. The heightened climate of intimidation eventually led to the decision of the Movement for Democratic Change to boycott the presidential run-off. A notable Zimbabwean scholar attributed the delay in the announcement of the presidential to the executive’s manipulation of the ZEC. 15 In Nigeria members of opposition parties and civil society leaders blamed INECs lack of independence during the 2007 elections for why candidates filed over 1,200 petitions to the Electoral Tribunal. The petitions filed in 2007 were more than double the number lodged after the 2003 Nigerian elections. Not only did the interlocutors consider EMBs important in their retrospective evaluations but also in their prognoses for the reforms that were necessary for the legitimacy of future elections. In South Africa, where the Independent Electoral Commission (IEC) has been a 14 A leading official of the National Patriotic Party (NPP), which lost the 2008 presidential elections mentioned that party officials and supporters were moderately confident in the electoral commission despite losing a very close race in the presidential elections (Personal Interview: Accra, May 2010). 15 Personal interview: Harare, Zimbabwe, August 2012. 171 forerunner in independent and professional management, leading political party representatives saw the body as being critical in addressing the problems of voter apathy through better voter education programs. In the opinion of the Chairwoman of the IEC, Dr. Brigalia Bam, the commission needed to make significant improvements in the voter education program as well as the selection process for ad-hoc staff so as to minimize allegations of political interference. 16 Most elites recognized the limitations that EMBs faced in guaranteeing credible elections. EMB commissioners and senior staff were most vocal in highlighting the ways in which political parties actively undermined the electoral commissions. According to a regional electoral commissioner in Ghana, “Political parties impair public confidence in the commission by spreading false allegations about the neutrality of the commission’s staff, while at the same time actively trying to bribe commission officials so that they will rig elections in their favor.” 17 More generally, elites admitted that the ability of EMBs to effectively organize elections is contingent upon the professionalism and neutrality of other state and political institutions such as security agencies, the judiciary, the bureaucracy, and the media. In commenting on the huge task of organizing elections in Zimbabwe, an electoral commissioner explained that “ZEC [Zimbabwe Electoral Commission] cannot work miracles . . . we have to recognize the role of the political parties and security agencies; this is especially the case because many of the security agencies are partisan.” 18 5.3.2 EMB Autonomy and Capacity 16 17 18 Personal interview, Pretoria, South Africa December 2010. Personal interview. Kumassi, Ghana May, 2010. Personal interview, Harare Zimbabwe, August 2012. 172 Across the four countries, elites regarded autonomy and capacity as distinct dimensions of EMB quality that had varying consequences on election quality. In both Ghana and South Africa the integrity and impartiality of the members of the national commission were indisputable. But elites in Ghana chastised the EC for capacity deficiencies in the voter registration process and in South Africa the IEC was criticized for the poor conditions of polling areas in depressed communities. By contrast, Zimbabwean elites lauded ZEC for its operational 19 capacity in 2008 (with the exception of finances and voter registration). Nevertheless, ZEC was considered blatantly partisan during the 2008 elections because Zanu-PF loyalist held important positions both in the secretariat and the commission. In Nigeria during the 2007 elections, elites bemoaned INEC’s lack of both autonomy and capacity. As a result, in the months before the 2011 elections many highlighted the dire need for reforms that could insulate the appointment process of INEC national commissioners and regional commissioners from executive control. They also expressed the need for an overhaul of the voter registration system, and for improvement in methods for selecting and training temporary poll workers. 5.3.3 Gap between EMB Design and Performance Most elites (even members of the ruling political party) saw the value in introducing formal reforms to enhance the autonomy and capacity of EMBs. 19 20 However, in Nigeria and A leader of a prominent civil society organization revealed that ZEC is significantly ahead of the other countries with respect to election logistics and planning. The main problem is with ad-hoc staff who are seconded from other ministries. Their training was inadequate during the 2008 elections. Due to the threat of violence there may be a shortage of qualified workers in the upcoming elections, leaving an opportunity for party activists to be ZEC staff (Personal interview: Harare, Zimbabwe, August, 2012). 20 As one Nigerian civil society leader explained “it is better to establish the rules and then hope one day that they will be enforced” (Personal interview, Abuja, Nigeria June 2010). 173 Zimbabwe most elites explained that there was a substantial gap between the statutory and constitutional changes that impact electoral management and the actual implementation of these rules. In particular, they highlighted the gap between the design and the actual performance of election administration. A first cause of the gap between design and performance was that politicians either lacked the will to implement the new rules, or else they undermined the rules through manipulation and fraud once they were implemented. In commenting on the effectiveness of the proposed constitutional amendments in Zimbabwe, a MDC-T Minster of Government explained that it depends on the extent to which political actors adhere to the reforms . . . within Africa there is a problem with the degree of institutionalization while it is easy to make laws there is need for greater conviction . . . without effective implementation free and fair elections will be impossible. 21 In the case of Nigeria, one prominent scholar admitted that the problem is not about the reforms themselves but the failure to adhere to them. “There is a culture of impunity in Nigeria, in which electoral offenders are not punished. This culture helped to undermine the efficacy of the 2006 Electoral Act.” 22 For most respondents, another cause for the gap was that EMB reforms were often passed very close to the date of the actual elections. Interviews were carried out in Nigeria (2010) and Zimbabwe (2012) during a constitutional reform process that included key changes to the legal framework governing elections in both countries. A common fear of civil society leaders and members of the donor community was that there would not be enough time for the reforms to be 21 22 Personal interview, Harare, Zimbabwe, August, 2012. Personal interview, Abuja, Nigeria June 2010. 174 effectively implemented in either country. In fact, the date of the elections in Nigeria had to be postponed on two occasions to facilitate the changes in the electoral timetable that were associated with these reforms. In the following section, I examine whether these elite perspectives on the link between EMBs and election quality across Ghana, Nigeria, South Africa, and Zimbabwe can be generalized to other countries in Africa. 5.4 Consequences of EMBs for Elite Attitudes and Behavior: A Statistical Analysis 5.4.1 Consequences for EMB Performance (De Facto Autonomy and Capacity) In this section, I explore the relationship between EMB performance (de facto autonomy and capacity) and four indicators of election quality: overall election quality, opposition participation, loser’s acceptance of results, and electoral turnover. The data used in this section involve information on 45 sub-Saharan countries for over 207 elections between 1990 and 2004 (See Table 5.1 for a description of the countries and elections). 23 I suggest that autonomous and competent EMBs increase the costs for political elites to engage in fraud and reduce the incidence of electoral irregularities. If the argument is accurate, we should expect EMB autonomy and capacity to be positively associated with overall election quality (H1), opposition participation (H2), losers’ acceptance of results (H3), and electoral turnover (H4). 5.4.2 Main Independent Variable: EMB Performance (De Facto autonomy and capacity) EMB performance, my main independent variable, is an indicator of the performance of electoral management bodies. It is constructed through an additive index of two variables that 23 For this analysis, I regard executive and legislative elections held concurrently as two separate elections. 175 reflect various attributes of EMB performance during the pre-election period and on election day. The variables are obtained from the Quality of Elections Data collected by Kelley (2012). Kelley codes these data from the U.S. State Department Human Rights Reports for the countries under investigation. I consider the EMB performance index a valid indicator of de facto autonomy and de facto capacity because the measure includes important evaluations of the actual autonomy and capacity of EMBs during various stages of the election cycle. According to Kelly, the measures reflect key aspects of EMB autonomy including the impartiality of the appointment process in practice or how evenly the EMB has been enforcing the electoral law. With respect to capacityrelated attributes, the measure captures logistical problems at voting locations, problems associated with election day workers, and deficiencies with the voter’s register. Like many cross-national measures of institutional characteristics, the EMB performance index does have its shortcomings. For instance, the index does not allow researchers to disaggregate the autonomy and capacity dimensions of the measure (which is an important aspect of the argument made in this chapter). Additionally, some may criticize the measure for focusing more on the capacity-related dimensions of the EMB performance. Despite these shortcomings, I am confident that the measure provides the most accurate and reliable gauge of EMB performance that allows for meaningful cross-time and cross-national comparisons. EMB performance index is normalized on a 0-1 scale with 1 indicating high levels of performance (see the Appendix for coding of all variables). variation across countries and within countries over time. 24 24 25 The measure shows substantial For instance, Ghana’s electoral I reversed the original coding of the variable to reflect increasing levels of de facto autonomy and capacity. 25 EMB performance index has a mean of 1.84, standard deviation of 0.199, minimum 0.166 and maximum of 1.0. 176 commission received a performance score of 0.5 during the 1992 elections; this increased to 1.0 between 1996 and 2000, but fell to .668 in the 2004 elections. In Zimbabwe the performance of election administration falls from 1 in 1990 to 0.67 in 1995 and down to 0.17 in 2000 but up to 0.5 in 2002. 5.4.3 Determinants of Overall Election Quality For the first assessment of election quality, I employ an indicator of overall election quality. Based on Hypotheses 1 (H1), I expect EMB performance to be positively correlated with overall election quality. To measure overall election quality, I rely on an indicator obtained from Lindberg’s Elections in Africa Data that measures whether the elections are judged essentially free and fair (2006). According to Lindberg, the variable captures whether electoral procedures were followed, whether people could join a political organization of their choice, whether demonstrations were allowed, and whether information was freely available. The variable is coded on a four-point scale 0-3, with 3 indicating that the election was entirely free and fair. 26 Between 1990 and 2004, 40 percent of elections were judged somewhat free and fair with irregularities affecting the outcome, while 50 percent of elections had irregularities that did not affect election quality. The remaining 9 percent of elections were either “entirely free or fair” (3%) or “not free and fair at all” (6%). As a first step in exploring the relationship between EMB performance and overall election quality, I consider the distribution of EMB performance scores across evaluations of election quality. As shown in Figure 5.1, 26 27 the mean EMB performance Four-point scale: 0 =“No” when no elections have been held or those held have been wholly unfair; 1 = “Irregular” when there were serious defects that influenced the results; 2 = “Yes, irregularities not significant” when there were deficiencies but they did not affect the result; and 3 = “Yes” when elections were free and fair. 27 All figures and tables are placed in the Appendix at the end of each chapter. 177 increases as the quality of elections improves: during election periods where elections are classified as not free and fair, mean EMB performance is 0.72. In contrast, EMB performance is 28 highest (1.0) in election periods that are classified as completely free and fair. With this in mind, I empirically test the relationship between EMB performance (independent variable) and overall election quality (dependent variable) while controlling for other relevant factors. 29 I incorporate the same controls used in the regression that examines the con- sequences of EMB design on elites’ election quality attitudes in Chapter 4 (Section 4). These controls include: proportionality of the electoral system and whether the system of government is presidential or parliamentary, level of fraud, electoral violence. I also consider the competitiveness of the elections, time since last elections, participation of international observers, and the level of democracy. I conduct a multiple regression analysis to examine the relationship between EMB performance and overall election quality, while accounting for confounding factors. 30 The results of the analyses are displayed in Model 1 of Table 5.2. The results provide strong confirmation for Hypothesis 1. That is, EMB performance is positively associated with election quality, even after controlling for other factors. We can use the change in predicted probability to provide a substantive interpretation of the results: increasing EMB performance from its 28 Furthermore, EMB performance has a moderate positive and significant bivariate correlation with overall election quality (Pearson correlation 0.392; p < 0.000). 29 See Appendix A1 for coding of variables. 30 Specifically, I utilize ordered logistic regression because it is the most appropriate method given that the dependent variable, election quality, contains four ordered categories. I also estimated the results using OLS regression and the results are not substantially different. 178 minimum to maximum value changes the predicted probability that elections will be free and fair by an average of 28%, holding all other variables at their means. 31 5.4.4 Determinants of Election Boycotts The existing literature and the findings from the elite interviews conducted in four African countries indicate that elites’ willingness to participate in elections may be influenced by the perceived autonomy and capacity of the EMB during the election period. While opposition parties often engage in strategic boycotts, there is still some reason to believe that lapses in the autonomy or capacity of election administration might so adversely undermine the integrity of elections that elites would boycott the elections (Beaulieu and Hyde 2009; Kelley 2012; Lindberg 2006; Bratton 1998). To empirically test the applicability of Hypotheses 2 (negative link between EMB performance and boycotts), I utilize a measure of opposition participation from Lindberg’s Quality of Elections Data. I dichotomize the measure with 0 indicating complete or partial boycotts by the main candidate (presidential elections) or by the main parties (parliamentarian elections) and 1 indicating full participation of non-incumbent parties or candidates. Across the sample of African elections, 1 in 3 elections experienced a complete or partial boycott. Examples include Benin during the 2001, Zambia during the 1996 elections, and Nigeria both in 1999 and 2003. As with the previous analysis, the main explanatory variable 31 In order to examine the robustness of the findings concerning EMB performance, I undertook the following analyses. First, I estimate additional regression analyses that include control variables that could influence overall election quality, including Log GNP, quality of previous elections, and whether or not the elections were founding multi-party elections. For all these analyses the significant and positive coefficient for EMB performance is maintained. Second, to account for the possibility of clustering across different countries, I ran the analysis in Model 1 in Table 5.2 with the standard errors clustered by country and the substantive results were not affected. Finally, I also conducted the analysis with an alternative dependent variables from Kelley’s Quality of Elections Database and the substantive findings were consistent with those in Model 1. 179 represents EMB performance (de facto autonomy/capacity) which provides an indication of the de facto autonomy and capacity of EMBs across Africa. Figure 5.2 highlights the variation in EMB performance across elections that experienced boycotts and those that did not. Although aver- age EMB performance is higher in elections where all political parties participated (0.84), the difference between this and elections in which boycotts occurred is not statistically significant. I conduct a multiple regression analysis to examine the relationship between EMB performance and election boycotts, while accounting for confounding factors. 32 The findings for the regression analysis are highlighted in Model 2 of Table 5.2. Contrary to the results regarding overall election quality, I find that EMB performance does not affect the likelihood of electoral participation across our sample. One explanation for this finding is that the decision to boycott may be driven by aspects of EMB performance in the pre-election period that may not be adequately represented in the measure of EMB performance. 5.4.5 Determinants of Losers’ Acceptance of Results Another way of understanding the effect of EMB performance on elite behavior is by assessing politicians’ willingness to accept the results of elections. Could it be that elites are more willing to accept defeat if the EMB displays high degrees of autonomy and capacity during the election? As highlighted in the introduction to this chapter, the opposition presidential candidate in Ghana’s 1992 elections attributed his party’s decision to challenge the results of the election to 1) flawed voters register and 2) blatant electoral fraud committed by INEC and the incumbent Rawlings administration. Past research also indicates that losing political parties and candidates are more likely to challenge election results when they have credible evidence 32 I utilize the same control variables used in Model 1 (Overall Election Quality). 180 indicating electoral fraud and procedural irregularities (Pastor 1999; Lindberg 2006). Although losers have a tendency to cry foul in the aftermath of an election, there is good reason to believe that on average, losers’ acceptance of results is positively associated with impartial and effective election administration, limited electoral fraud and manipulation, widespread freedom to campaign, and equitable electoral rules (Bratton 1998). To examine these conjectures, I utilize an indicator that captures the extent to which losers accept the results of elections (Lindberg 2006). The variable is coded from 0 to 2, with 2 indicating that all main players accepted the results. Across the elections included in the data, approximately 1 in 3 elections saw all the political actors accepting the election results, while in the remaining 71 percent of elections either none of the parties accepted the results (34%) or only a proportion of the parties accepted election outcomes (37%). The findings from the bivariate analysis show that average EMB performance is highest in election periods where political elites accept the results (0.92). Meanwhile performance is lowest when results are not accepted by all of the main parties (0.79) (See Figure 5.3). I also estimate a multiple regression analysis (Model 3 Table 5.2). The results clearly highlight the significant and positive coefficient for EMB performance. I interpret this to mean that during elections where EMBs display high levels of political independence and efficiency in election administration, losers are more willing to accept the results of elections. Political parties are 24 percent more likely to accept the results of elections (losers’ acceptance of results = 2) when we increase EMB performance from its minimum to maximum value. 5.4.6 Determinants of Electoral Turnover Many students of democracy have identified alternations of power as a sufficient condition for the consolidation of democracy (Huntington 1991; Przeworski 2000). For this 181 reason, I also examine whether the quality of election administration has any significant relationship with political turnover in Africa. I suggest that more politically insulated and administratively competent EMBs will more likely lead to turnovers in power, because it makes it more difficult for the incumbent to rig elections. Many scholars have associated Ghana’s achievement of the “two-turnover” test with the independence and professionalism of the Electoral Commission. However, we only need look to the example of South Africa, a country that has not experienced the electoral success of an opposition party in national elections, to see that EMB performance is not a sufficient condition for electoral turnover. The main dependent variable in this analysis is electoral turnover, coded from 0 to 2, where 2 signifies complete turnover in power (Lindberg 2006). Among the 45 countries included in the sample, complete alternation of power only occurred in 20 percent of elections, partial turnovers in 10 percent, while the remaining 70 percent of elections ruling parties won elections. I first show the bivariate relationship between electoral turnover and EMB performance. Notice in Figure 5.4 that mean EMB performance is highest in elections that result in complete alternation of power (turnover = 2: 0.92) and lowest in elections where the incumbent retains office (turnover = 0: 0.79). What relationship, if any, does alternation of power have with EMB performance in a multiple regression model? Model 4 of Table 5.2 summarizes the results of the multiple regression analysis. The results reinforce the important role of EMB performance: during election periods that EMBs perform effectively, political leadership will more likely change hands from one party to another. 5.5 Discussion We can infer from findings in Models 1-4 that EMB performance is an important correlate of overall election quality, losers’ acceptance of results, and turnover in elections, even 182 after including relevant controls. This is a significant finding for the following reasons. First, from a theoretical standpoint, the findings provide strong support of my argument that election administrative performance matters. EMB performance enhances the quality of elections by providing incentives for political actors to adhere to the rules governing elections and by reducing the extent to which procedural irregularities occur. Moreover, the results also show that EMBs perform best in emerging democracies or hybrid regimes. Second, the results confirm many of the existing comparative studies that identify EMBs in Africa as vital institutions in the democratization process (Ayee 1998; Debrah 2011; Masunungure 2009; Trebilcock and Chitalkar 2009). Third, from a methodological perspective, the results demonstrate that the indicator of EMB performance (Kelley 2012) provides a reasonably effective indicator of election administrative performance that can be utilized in cross-national and cross-time comparisons. Moreover, it allows researchers to move beyond the use of formal-legal measures to behavioral indicator that more directly taps into institutional performance. One potential criticism of the findings is that EMB performance is endogenous to election quality: perhaps these measures of election quality, especially overall election quality assessments, actually account for the performance of election administrative bodies. In the research design, I have taken two steps to address this issue of endogeneity. First, for the four analyses, the main dependent variables (overall election quality, boycotts, losers’ acceptance of results and turnover) and our main independent variable (EMB performance) were obtained from different data sources. This allows for a certain degree of independence in the coding of these variables. Second, by examining the relationship between EMB performance and two more objective indicators (turnover and losers’ acceptance of results) other than overall election quality assessments, I am able to demonstrate the robustness of the effect. While one could make 183 a reasonable argument that the measure of overall election quality (Lindberg, 2006) includes experts’ assessment of EMB performance, this is less so for losers’ acceptance of results and electoral turnover. 5.6 Conclusion Do EMB autonomy and capacity matter for political elites in Africa? By using a mixed method approach which incorporates data from elite interviews in Ghana, Nigeria, South Africa and Zimbabwe along with a cross-national statistical analysis between 1990 and 2004, this chapter provides strong evidence to show that EMB autonomy and capacity are importantly associated with elite attitudes and behavior during elections. I find that the actual performance of EMBs (de facto autonomy and capacity) during elections has an important effect on shaping the incentives and behaviors of elites. Elites are more likely to play by the rules of the game, participate in elections, and accept results when EMBs display high levels of autonomy and capacity. Moreover, this relationship is more likely to occur in emerging democracies with weak rule of law and limited restrictions on civil liberties. In the following chapter I shift the spotlight from elites to see how citizens’ attitudes towards elections are influenced by the autonomy and capacity of EMBs in Africa. Here I find that citizens’ attitudes towards elections are strongly and positively associated with the performance of EMBs. 184 APPENDIX 185 Figure 5.1: Mean EMB Performance by Overall Election Quality 1.00 1.00 0.88 0.90 0.80 0.72 0.77 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 Not Free and Fair Significant Limited Free and fair (7) (12) Irregularities (86) Irregulariites (108) 186 Figure 5.2: Mean EMB Performance by Opposition Participation 0.90 0.84 0.81 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 Opposition Boycott (65) Opposition Particiaption (148) Mean EMB Performance 187 Figure 5.3: Mean EMB Performance by Losers’ Acceptance of Results 1.00 0.92 0.90 0.80 0.79 0.80 0.70 0.60 0.50 0.40 No Main Losers Accepts Results (73) Some Losers Accept Results (79) Mean EMB Performance 188 All Losers Accept Results (61) Figure 5.4: Mean EMB Performance by Electoral Turnover 0.94 0.92 0.92 0.89 0.90 0.88 0.86 0.84 0.82 0.80 0.80 0.78 0.76 0.74 0.72 No turnover (151) Partial turnover (20) Mean EMB Performance 189 Complete Turnover (42) Table 5.1: Country and Election Years- 45 Country Analysis Country Freq. Algeria 6 Benin 7 Botswana 3 Burkina Faso 5 Burundi 2 C. Verde Is. 6 Cameroon 6 Cent. Af. Rep. 6 Chad 5 Comoro Is. 8 Congo 7 Cote d’Ivoire 6 Djibouti 5 Egypt 4 Eq. Guinea 5 Ethiopia 4 Gabon 5 Gambia 5 Ghana 8 Guinea 6 Guinea-Bissau 5 Kenya 6 Lesotho 3 Election Years 1991, 1995, 1997, 1999, 2002, 2004 1991 (L/E), 1995, 1996, 1999, 2001, 2003 1994, 1999, 2004 1991, 1992, 1997, 1998, 2002 1993 (L/E) 1991 (L/E), 1995, 1996, 2001 1992 (L/E), 1997 (L/E), 2002, 2004 1992 (L/E), 1992 (L/E) 1998, 1999 1990, 1996, 1997, 2001, 2002 1990 (L/E), 1992, 1993, 1996, 2002, 2004 1993,2002 1992, 1990 (L/E), 1995 (L/E), 2000 (L/E) 1992, 1993, 1997, 1999, 2003 1990, 1993,1995,2000 1993, 1996, 1999, 2002, 2004 1992, 1994, 1995, 2000 1990, 1993, 1996, 1998, 2001 1992, 1996, 1997, 2001, 2002 1992 (L/E), 1996, 2000, 2004 1993, 1995, 1998, 2002, 2003 1994 (L/E), 1999 (L/E), 2004 1992 (L/E), 1997 (L/E), 2002 (L/E) 1993, 1998, 2002 Country Freq. Liberia 2 Madagascar 6 Malawi 7 Mali 7 Mauritania 6 Mauritius 3 Mozambique 6 Namibia 6 Niger 11 Nigeria 7 Rwanda 2 S. Africa 3 Senegal 5 Sierra Leone 4 Sudan 4 Swaziland 3 Tanzania 5 Togo 9 Tunisia 3 Uganda 5 Zambia 6 Zimbabwe 6 190 Election Years 1997 (L/E) 1992, 1993,1996, 1998, 2001, 2002 1992, 1994(L/E), 1999 (L/E), 2004 (L/E) (L/E), 1997 (L/E), 2002 (L/E) 1992 1992 (L/E), 1996, 1997, 2001, 2003 1991, 1995, 2000, 1994 (L/E), 1999 (L/E), 2004 (L/E) 1994 (L/E), 1999 (L/E), 2004 (L/E) 1993 (L/E) 1995, 1996, 1999, 2004 1992, 1993, 1998, 1999(L/E), 2003(L/E) 2003 (L/E) 1994, 1999, 2004 1993 (L/E), 1998, 2000, 2001 1996 (L/E), 2002 (L/E) 1996 (L/E), 2000 (L/E) 1993, 1998, 2003 1990, 1995 (L/E), 2000 (L/E) 1990, 1993, 1994, 1998, 1999, 2002, 2003 1999, 2004 1994, 1992, 1996(L/E), 2001 (L/E) 1991(L/E), 1996 (L/E), 2001 (L/E) 1990, 1995, 1996, 2000, 2002 Table 5.2: Correlates of Election Quality (45-country sample) Variable EMB Performance (de facto autonomy/capacity) Proportional Electoral Rules Parliamentary Elections Election Fraud Election Violence Competitiveness of Elections International Observers Freedom House Index (Lag) Time Since Last Election Pseudo R2 Log Likelihood N Model 1 Overall Election Quality Ordered Logit Model 2 Opposition Participation Logit Model 3 Losers’ Acceptance of Results Ordered Logit Model 4 Electoral Turnover Logit 3.061*** -0.907 1.522* 2.799** (0.966) 0.00213 (0.228) 2.077** (0.817) -3.627*** (0.762) -1.130*** (0.342) -0.00733 (0.00664) -0.232 (0.397) 3.656*** (1.110) 0.144* (0.0750) 0.394 -104.680 185 (0.983) 0.337 (0.268) 1.468 (1.145) -0.844 (0.746) -0.968*** (0.362) -0.0353*** (0.00744) 0.457 (0.433) 0.211 (1.187) 0.225** (0.0877) 0.274 -79.102 185 (1.238) -0.150 (0.190) -0.00719 (0.621) -1.774*** (0.573) -0.348 (0.271) 0.0111* (0.00592) 0.178 (0.325) 0.701 (0.827) -0.0612 (0.0616) 0.176 -166.30 185 (1.238) -0.166 (0.248) -0.741 (0.806) -5.180*** (0.972) 0.320 (0.338) --0.217 (0.431) -1.188 (1.060) -0.223** (0.0893) 0.322 -76.044 189 191 CHAPTER 6 THE CONSEQUENCES OF EMB PERFORMANCE ON CITIZENS’ ELECTION QUALITY EVALUATIONS 6.1 Introduction: Give us a Competent and Unbiased Electoral Umpire! On March 10th of 2010, thousands of Nigerians marched through the capital of Abuja bearing placards that indicated their disgust with the country’s electoral management body: the In- dependent National Electoral Commission (INEC). 1 Nigerians from all walks of life demanded, among other things, the immediate sacking of the Chairman of INEC, Professor Maurice Iwu, for his mismanagement of the 2007 elections and his perceived allegiance to the ruling Peoples Democratic Party. Protestors also called for the implementation of the Justice Uwais Panel Report; which advocated a series of electoral reforms following the 2007 Nigerian elections. 2 These mass protests for electoral reform in Nigeria provide evidence of a general trend in Africa whereby everyday citizens demand the implementation of policies to enhance the autonomy and administrative efficiency of electoral management bodies. Recently, citizens in Ivory Coast, Kenya, Lesotho, Malawi, Zanzibar, Zimbabwe, and Zambia have staged public demonstrations to express frustration with electoral commissions perceived as politically manipulable, complicit in electoral fraud, and ill prepared for elections. In some cases, citizens 1 Nigerians were also concerned with the constitutional crisis associated with the prolonged illness of the country’s president Umaru Musa Yar’Adua. For more information on the March 2010 protests see Obasi (2010) and LeVan and Ukata (2012). 2 The Electoral Reform Committee (ERC) was headed by former Chief Justice Muhammadu Uwais. The ERC produced a report that reflected nationwide consultations with stakeholders on how to improve the electoral process before the 2011 elections. 192 insisted on changes in the legal framework, such as the process of appointing EMB commissioners. In other cases, citizens demanded more credible voter’s list, or called for the removal of EMB commissioners. In this chapter, I shift the focus from elites to citizens. Specifically, I examine how citizens formulate their opinions of election quality in Africa and consider whether the autonomy 3 and capacity of EMBs influences citizens’ election quality opinions. I conceptualize election quality as citizens’ evaluations of the freeness and fairness of all stages in the processes of 4 elections (pre-election period, election-day, post-election period). As in the previous chapter, I distinguish between two interrelated EMB characteristics: those related to the 1) autonomy and 2) capacity of EMBs. Exploring ways in which citizens formulate their perceptions of election quality in Africa is important for a number of reasons. First, various studies emphasize the strong association between citizens’ election quality perceptions and citizens’ democratic attitudes. When citizens consider elections to be free and fair, they are more likely to espouse positive evaluations of democratic performance and to believe in the legitimacy of the regime (Bratton 2007; Moehler 2009; Chu et al. 2008; Birch 2011). Popular election quality attitudes are also important predictors of citizens’ electoral behavior. Not only do instances of voter intimidation, violence, and ballot box stuffing drive citizens to protest but they also enhance the likelihood of postelection violence (such as in the case of Kenya in 2007) and regime collapse (such as the case of 3 It is also possible that EMBs can influence citizens’ electoral behavior as seen clearly with the example of the protests staged in different African countries; however, examining the impact of EMBs on political behavior is outside the scope of this analysis. 4 As I will explain in more detail below, this conceptualization of election quality from the perspective of everyday citizens is consistent with the main notions of election quality highlighted in the existing literature. For more information see Norris 2012. 193 5 Ivory Coast in 2010). Third, many scholars suggest that the perspectives of citizens gathered through public opinion surveys (based on representative samples), provide a more representative snap- shot of election quality within a country than do elite and expert opinions of election quality (Norris 2012; Banducci and Karp 2003). Despite the importance of gauging citizens’ election quality perceptions, the existing literature has been relatively quiet on how Africans’ formulate their perceptions of election quality. Some scholars have conducted cross-national studies that systematically examined 6 popular perceptions quality in other developing regions of the world. But, while other scholars have employed single-case studies in the context of African countries, 7 no studies, to my knowledge, have investigated the causes of mass election quality perceptions using crossnational data on elections in Africa. In this chapter, I contend that citizens’ election quality perceptions are influenced by multiple factors, including the design and performance of institutions related to electoral administration, characteristics of the election environment, and citizens’ individual attributes. Of principal importance is understanding whether EMB performance shape citizens’ attitudes towards election quality in Africa. The chapter proposes a theoretical model, based on historical institutionalism, to explain the possible effect of EMB autonomy and capacity on citizens’ election quality perceptions. I suggest that Africans have become increasingly critical of the quality of elections (Bratton 2013: 5 For more studies that explore the effect of election quality judgements on citizens’ attitudes see: Lindberg 2006, 2009; Robbins & Tessler 2010. For studies that examine the consequences on political behavior see McCann & Dom´ınguez, 1998, Birch 2011, Bratton 2008, Collier and Vicente 2009, Kuntz and Thompson 2009, Dercon and Gutierrez-Romero 2011, Eisenstadt 1999, 2002, Pastor 1999, and Schedler 2002. 6 For instance, see Rosas 2010 in Latin America. 7 See, for instance, Adebayo and Omotola (2007) on public perceptions of the 2007 Nigerian elections. 194 31). They want elections in which the playing field is even, voting is hassle free, and votes are accurately counted. EMBs become crucial institutions for citizens during elections because EMBs directly affect the extent to which politicians adhere to the rules of electoral competition 8 as well as the pervasiveness of problems with the administration of elections. This argument is an extension of the one proposed in Chapter 5, which links EMB autonomy and capacity with politicians’ incentives to engage in fraud, participate in elections, and accept defeat. In this chapter I show that citizens’ confidence in the freeness and fairness of elections is boosted directly by experiencing fewer instances of electoral fraud and lapses in electoral administration, or indirectly as they are exposed to fewer complaints by political elites (civil society and the media) about electoral fraud and problems with electoral administration (See Figure 1.2). In the following section (Section 6.2), I survey the existing literature on various factors that shape citizens’ election quality perceptions. In Section 6.3, I highlight pathways through which EMBs potentially shape citizens’ perceptions of election quality. Importantly, I show how both the autonomy and capacity of EMBs can create a perception of a more equitable electoral playing field and minimal administrative irregularities. I also suggest that citizens are more likely to evaluate EMBs based on aspects of its performance (de facto autonomy/capacity). I rely on data from the Afrobarometer public opinion survey to describe Africans’ perceptions of election quality as well as their evaluations of election administration between 1999 and 2008 across 18 countries (Section 6.4). In Section 6.5, I statistically examine the association between citizens’ perceptions of election quality (the dependent variable) and EMB autonomy and capacity (the 8 As will be explained in the following section, I distinguish between electoral manipulation and electoral administrative irregularities. Electoral administrative irregularities are conceptualized as any systematic deficiency in election administration that is not a consequence of intentional manipulation. 195 independent variables). 9 The empirical findings indicate that Africans’ perceptions of election quality are positively related to EMB performance (de facto autonomy/capacity). The conclusion (Section 6.6) briefly summarizes of the findings. 6.2 Literature Review 6.2.1 Conceptualizing Election Quality Perceptions Citizens’ overall assessments of election quality reflect, to some degree, popular understandings of the freedom and fairness of various stages of electoral processes (Dahl 1971; Svensson and Elklit 1997; Goodwin-Gill 1994; Massicotte 2005). “Freedom” connotes candidates’ ability to run for elected office, and engage in campaigning, as well as the voters’ ability to choose a preferred outcome, all without constraint or coercion. “Fairness” signifies the impartial and consistent application of rules (international and domestic) in ways that do not adversely bias some political competitors relative to others. The fairness dimension is often evidenced through impartial and effective conduct of electoral-related institutions such as the electoral commission, security forces, media and courts and the absence of systematic electoral fraud and manipulation. More recent studies have also associated election quality with concepts of electoral integrity and electoral malpractice, but these more recent conceptualizations encompass similar processes through which citizens develop their opinions of election quality (Norris 2012; Birch 2011). 10 The scope of citizens’ election quality perceptions is also important. 9 Similar to Chapter 4, I rely on an original data set on EMB design across 20 African countries between 1990 to 2008. To gauge the opinions of citizens, I incorporate survey data from the Afrobarometer public opinion project. 10 According to Norris (2012), electoral integrity denotes “shared international principles, values, and standards of democratic elections which apply universally to all countries and which should be reflected at all stages during the electoral cycle, including the pre-electoral period, the 196 Studies assume that citizens’ evaluations are not limited to the activities on election day; instead they span all the stages of the electoral process, starting from the designing of legislation related to election administration to the adjudication of electoral disputes. 11 6.2.2 Sources of Citizens Election Quality Perceptions The emerging literature on mass perceptions of election quality highlights numerous factors that shape citizens’ views on the freeness and fairness of elections. For the purpose of this analysis, I group these potential factors into three broad categories: the performance of institutions associated with election administration (e.g., electoral management bodies), characteristics of the electoral environment (e.g., electoral manipulation) and citizens’ individual attributes (e.g. political preferences and political knowledge). 6.2.3 Institutional Performance In Chapter 4 (Section 2), I outlined the existing literature on how EMB design structures citizens’ election quality attitudes. While investigating the relationship between formal rules of election administration and citizens’ attitudes toward elections is important, a principal limitation is that formal-rules may not adequately reflect the actual performance of EMBs (Bratton, Mattes and Gyimah-Boadi 2005; Mattes and Shenga 2013). 12 To overcome this problem, scholars have begun to explore the link between the performance of institutions such as electoral management campaign, polling day and its aftermath.” Electoral malpractice, on the other hand, refers to attempts to violate electoral integrity. 11 For more detail on the stages of the electoral process see Norris 2012, and Birch 2011. 12 Similar to Chapter 4, we distinguish between formal-legal (de jure) and performancebased (de facto) dimensions of EMB autonomy and capacity. Scholars using cross-national data on perceptions of election quality including Birch (2008) and Rosas (2010:88) note their inability to evaluate EMB performance directly because of the unavailability of reliable cross-national data. However, they suggest that utilizing performance based-measures would provide greater understanding of the consequences of EMBs. 197 bodies and citizens’ election quality evaluations. Most studies have sought to explain the impact of EMB performance on citizens’ election quality perceptions using within country cross-time and cross-sectional analyses. Take for example the innovative work by Hall et al. (2009). The authors examine the competence of election workers in the states of Ohio and Utah (an attribute of EMB administrative capacity) and find that poll worker performance enhances citizens’ confidence in election legitimacy. 13 There have also been numerous studies in Africa that highlight the dynamics of EMBs influence on citizens’ attitudes and behaviors (Gyimah-Boadi 1999; Adebayo and Omotola 2007; Debrah 2011). Focusing on the performance of the Electoral Commission (EC) in Ghana between 1992 and 2008, Debrah clearly highlights how the EC was able to demonstrate to election stakeholders, its ability to preserve its constitutionally enshrined independence by withstanding pressures from successive political parties and effectively manage the electoral process. Conversely, Adebayo and Omotola (2007) reflecting on the 2007 Nigerian elections find that systematic lapses in the administration of elections by the Independent National Electoral Commission (INEC) (such as the late or non-opening of polling stations and inaccurate voters roll) as well as the collusion between INEC officials and political actors to conduct ballot fraud, were the main reasons why the public had little or no confidence in the elections. 6.2.4 Characteristics of the Electoral Environment Despite the disproportionate focus on institutional sources of citizens’ opinions of election quality, there are other factors specific to the electoral environment that are influential. The literature often points to citizens’ experience with electoral manipulation as a principal 13 For other studies on EMB performance in the United States see: Atkeson et al 2007; Alvarez et al 2008; Burden et al 2011; Classen et al, 2008, 2012; Hall, 2003; Hall et al, 2009; Hale and Slaton 2008. 198 source of their election quality opinions. 14 Acts of intimidation, violence, ballot rigging, and vote buying all challenge popular notions of electoral equity because they provide substantial advantages for the perpetrators at the expense of other competitors (Schedler 2006). 15 Furthermore, when such irregularities become pervasive during the election period they provide a disincentive for voter participation and undermine the quality of the voting experience (Bratton 2008). Because of this, voters may think that freeness and fairness of the election has been compromised. Existing cross-national studies that explore sources of election-quality perceptions have encountered difficulties in integrating direct indicators of electoral irregularities within their analysis. For instance, Birch (2008) uses an aggregate measure of corruption perceptions— the Corruption Perceptions Index (CPI)—as a proxy for the level of fraud. She assumes that in countries where political corruption is pervasive, citizens are more likely to consider the administration of elections to be corrupt and by extension consider elections as falling below democratic standards. Another group of studies rely on micro-level data to assess citizens’ perceptions of, and experience with, electoral irregularities (Schedler 1999; Alvarez et al., 2008a; 14 For the purposes of this chapter, I define electoral manipulation as the range of illicit activities that political actors employ to alter election outcomes or to undermine the electoral process (Lehoucq 2003). Some of the most common examples of electoral manipulation include vote buying, electoral violence and intimidation, ballot box stuffing and falsification of results etc. For a more systematic analysis of electoral manipulation see Schedler 2002; Lehoucq 2003; Alvarez et al 2008; Frazer and Gyimah-Boadi 2012. Importantly, I distinguish between electoral manipulation and electoral administrative irregularities. Electoral administrative irregularities are conceptualized as any systematic deficiency in election administration that is not a consequence of intentional manipulation. 15 Some readers may find the distinction between citizens’ experience with electoral manipulation from their perceptions of election quality ambiguous. While the existing literature associates the prevalence of electoral fraud with low levels of electoral quality or electoral integrity, I attempt to test this association empirically by examining the ways in which aggregate levels of electoral fraud influence citizens understanding of election quality. 199 Bratton 2008; Collier and Vicente 2009). Bratton (2008) examines citizens’ perceptions of electoral irregularities in the context of the 2007 Nigerian election campaign period. Nigerians who experience threats of violence are less likely to vote in the 2007 elections or support the ruling party. Those offered material benefits in exchange for their votes express greater partisan loyalty; but they are no more likely to participate in elections than those not offered these material benefits. 16 6.2.5 Citizens’ Individual Attributes A final set of sources of citizens’ understandings of the freeness and fairness of electoral contests is their own individual attributes; these include partisan affiliations, political knowledge, and political engagement. The importance of political preferences as a source of election quality perceptions is often emphasized within the literature (Alvarez et al. 2008b; Moehler 2009; Rose and Mishler 2009). Citizens’ political preference is represented as either an ingrained ideological attachment to a political party or an affiliation with the winning or losing party in the most recent elections (Anderson and LoTempio 2002; Bratton, Mattes and Gyimah-Boadi 2005). Regardless of how citizens are affiliated with political parties, these attachments structure their perceptions of election quality: those affiliated with the winning party are more inclined to believe that the success of their party was legitimate, while those affiliated with the losing party search for 16 Although existing studies directly examine citizens’ experience with electoral irregularities and deal with the consequences on electoral behavior, none of them directly test the effect of these experiences on citizens’ perceptions of election quality. Nevertheless, these studies are instructive because they identify various potential causal pathways between experience with election irregularities and evaluations of election quality. For instance, if we build upon Brat- ton’s (2008) findings regarding the impact of intimidation we would expect that citizens who experience irregularities may score the elections poorly not only because of their experience but because they were also deprived of their constitutional right to participate in the elections. In sum, by studying various forms of electoral irregularities, we can gain a better understanding of how citizens respond to elite strategies to manipulate the electoral process. 200 evidence to show that their favored candidate or party was short- changed. 17 Moreover, political parties represent an important source of information on the quality of elections (Lassen 2005). In a recent cross-national analysis of citizens’ perceptions of election quality in Africa, Moehler (2009:16) considers affiliation with the electoral winner. She finds that, relative to all other factors, “winner status has the largest influence on evaluations of electoral integrity in both substantive and statistical terms.” 18 Another important individual attribute is citizens’ level of knowledge about election processes. 19 Citizens’ level of knowledge about electoral processes, or more generally their 20 political knowledge, could potentially influence the formation of election quality perceptions. More specifically, I suggest that politically informed citizens have greater insight into the 17 Partisanship also influences citizens’ views on the freeness and fairness of the electoral contest. Once election results have been announced, a citizen’s status as the winner, loser, or non-partisan influences his/her election quality evaluations through two main mechanisms. The first relates to the citizens’ interpretation of election results whereby citizens’ affective or instrumental association with the electoral winner (loser) drives their satisfaction (disappointment) with the election. A second pathway relies more on the perceptions of freeness and fairness of the electoral contest. Here partisanship provides a schema through which citizens assess the impact of institutional performance and election irregularities. As mentioned before, losers are more likely than winners to identify performance deficiencies, electoral manipulation, and other irregularities as justification for their candidate’s loss. 18 Similar research by Alvarez et al. (2008b) shows that, within the American context, partisanship has a significant impact on citizens’ confidence in the accuracy of the vote count and tabulation for the US presidential elections in 2000 and 2004. Republicans, compared to Independents and Democrats, were most likely to display confidence in the elections on the grounds that their candidate won. 19 The literature on political knowledge is too extensive too record here; for an exhaustive review see Norris 2011 (142-187). 20 The existing literature is divided on the extent to which citizens can be expected to understand the intricacies of elections (Banducci and Karp, 2003; Birch, 2008; Pastor, 1999; McCann and Dominguez, 1998; Bratton et al., 2005; Conroy-Krutz, 2009). Without taking a specific position on this debate, I contend that knowledge about electoral processes will vary among citizens, with some citizens having greater information about electoral processes than others. 201 21 principles that guide free and fair elections. As a result, they are more likely to recognize when elections adhere to, or deviate from, these principles. The existing literature has highlighted this dynamic from two vantage points. One group of studies conducted in developed democracies has found that informed citizens are more inclined to have positive evaluations of election quality (Alvarez et al. 2008; Birch 2008; Norris 2004). The main rationale for this is that informed citizens are more likely to participate in elections. And relatively higher degrees of participation increase citizens’ familiarity with election processes, which in turn enhances confidence in elections. However, scholars are keen to point out that when elections do fall below acceptable standards those who participate are usually the ones more willing to cry foul. A second group of studies indicates a negative association between political knowledge and election quality attitudes in Africa (Moehler 2009). In particular, Moehler’s study shows that Africans who rely frequently on the radio for political information (as well as those with higher levels of education) were more critical of election quality across 18 African countries. 22 Bratton et al. (2005) support this perspective. Here the authors demonstrate that citizens with higher levels of media exposure tend to be more critical of the performance of government. This is because these citizens are more aware of the role governments should play in democracies. If we extend this rationale to our assessment of election quality, we may expect politically informed 21 Citizens can gain information about elections processes directly or indirectly. Information gathered through direct experience is more reliable, but limited in its scope. While indirect sources of information provide a more representative perspective of election processes. However, taking advantage of this depends on the availability of information; the accuracy of information; and, citizens’ willingness to seek out information on elections. 22 This finding is line research by Bratton et al. (2005) that shows how citizens with higher levels of media exposure tend to be more critical of the performance of government. This is because these citizens are more aware of the role governments should play in democracies. If we extend this rationale to our assessment of election quality, we may expect media-savvy citizens to have greater insight into what free and fair elections should be, and more likely to recognize when elections deviate from these norms. 202 citizens to have greater insight into what free and fair elections should be, and more likely to recognize when elections deviate from these norms. 6.3 How Do Citizens Evaluate Election Quality in Africa? 6.3.1 What Types of Elections Do Citizens Want? I start out by assuming that citizens desire elections that are free and fair. Elections are “instruments of democracy” (Powell 2000), and free and fair elections provide the best conditions for decisions at the ballot box to be effectively translated into the selection of a candidate (or political party) who can most effectively pursue citizens’ preferred policies. This is not to say that some citizens will not accept fraudulent elections if they usher their favored candidate (party) into office. But, I expect that most citizens will reject elections outright that are perceived as being unfair when they prevent the victory of the candidate (party) of their choice. More specifically, however, citizens demand elections in which the electoral playing field is level. That is, there is equal access to campaign resources among candidates and there are sufficient checks on incumbent manipulation and fraud. Citizens also demand elections that have as few administrative problems as possible. The first reason for this is that citizens want to be confident that their votes are accurately counted. Second, citizens want a voting experience that is as hassle free as possible and thereby minimize the costs of voting (Downs 1957). 6.3.2 How do EMBs Help Promote Free and Fair Elections? Africans are not only clamoring for free and fair elections, but they are also becoming more aware of the role that EMBs serve in promoting credible elections. 23 The growing importance of EMBs is associated with greater experience with multiparty elections in which 23 In Section 4.3, I present data that shows how citizen awareness of EMBs in 12 African countries has been increasing over time since 1999. 203 domestic and international actors have stressed the significance of EMBs. 24 With this growing public awareness, citizens have also begun to develop expectations of what EMBs should do to effectively manage elections. For many citizens, the role of the EMB is analogous to an umpire/referee in competitive team sports such as football or cricket. 25 Citizens expect the umpire to be impartial and competent in managing both teams and always willing to enforce the rules of the game when violated. If we extend this analogy to EMBs, the first duty of the organization is to ensure that the electoral playing field is level. This can be achieved through a variety of ways, including sanctioning political parties for violating electoral rules and international norms; making decisions that are transparent and insulated from political interference, and engaging all political parties and promoting inter-party cooperation. A second duty of EMBs is to ensure that election-related irregularities are minimized. This can range from EMB officials receiving the necessary training on election procedures and changes in election technology to EMBs providing the necessary resources (such as ballot papers, indelible ink, and ballot boxes) on polling day to ensure that all registered voters get the opportunity to cast their ballots. In analyzing the direct effect of EMBs on citizens’ election quality perceptions there are two questions that should be considered. First, what characteristics of the EMB are most important when citizens evaluate its effectiveness in promoting fair and 24 Even beyond the boundaries of their own country, Africans have been exposed to the success of electoral commissions in countries like Ghana and South Africa and how these commissions have been instrumental to guaranteeing democratic consolidation. While on the other hand, the infamy of commissions in Kenya (2007) and Zimbabwe (2005) has served as important reminders as to how EMBs can precipitate democratic instability and political conflict. 25 Through personal conversations with citizens and elites in Nigeria (2010 and 2011) and Ghana (2008 and 2010) the electoral commissions were often referred to as electoral umpires: and the analogy with sports umpires was frequently made. 204 well-organized elections? And, second, what are the sources of information that citizens rely upon to evaluate EMBs? 6.3.3 EMB Characteristics: Autonomy and Capacity There are various characteristics of EMBs that citizens may consider important when evaluating the institutions’ ability to reduce electoral fraud and minimize electoral irregularities. 26 I choose to focus on two interrelated dimensions of EMB characteristics: those related to the 1) autonomy and, 2) capacity of EMBs. One reason for making this distinction is that the characteristics related to EMB autonomy may have a different effect on citizens’ election quality perceptions when compared to the characteristics related to EMB capacity. For instance, the perception that the final results of an election were doctored by the EMB in favor of the incumbent (allegations of this sort were made by various stakeholders of the Zimbabwean Electoral Commission following the March, 2008 national elections) may impact citizens’ election quality evaluations differently from the failed implementation of an electronic voter registration system (as was the case in Nigeria during the 2007 elections). 27 Furthermore, a growing number of theoretical, empirical, and policy-related studies on electoral governance have also underscored the importance of distinguishing between these two dimensions (Rosas 2010: IDEA 2012). 28 Finally, and most germane to the main argument, is the idea that some citizens have also been making this distinction when assessing EMBs. 26 These characteristics are exhaustive, including impartiality, administrative efficiency, transparency, public accountability and poll-worker competence. 27 For more information on the allegations of fraud associated with ZEC see Ploch (2008: 6) 28 In fact the policy community has been most consistent in making this distinction between autonomy and capacity. International donor organizations, international observers, and domestic organizations have promoted distinct policies to enhance both the capacity and autonomy of EMBs. 205 6.3.4 EMB Autonomy Autonomy is important for citizens because it provides information on the ability of EMBs to resist incumbent manipulation and deter electoral fraud. To put it plainly, citizens’ attitudes on electoral freeness and fairness are affected by the perceived neutrality of the electoral umpire and the extent to which the umpire will ensure that all political actors play by the rules of the game. As a consequence, when EMBs are perceived to be in the back pocket of the incumbent, citizens are less likely to have confidence in the institution’s ability to carry out its functions during the course of the elections. Most importantly, citizens will question whether the final results of the elections actually represent the collective will of voters, because EMB officials play such a critical role in voting operations. As highlighted in previous chapters, there are numerous attributes of EMB autonomy. From a citizen’s perspective, one very important dimensions of autonomy is the impartiality of EMB executives and staff. According to the literature on bureaucratic agencies, citizens may either focus on the reputation of bureaucratic executives (who are highlighted in popular media) or the street-level bureaucrats they come in contact with on a regular basis (Elklit and Reynolds, 2005). I extend this insight to the analysis of EMBs and argue that citizens are more likely to focus on attributes that influence the selection and conduct of EMB commissioners, regional executives, as well as EMB staff (especially those temporary polling day workers). For instance, citizens tend to associate the process of appointing the EMB chairperson (and reputation of the chairperson) with the organizations’ ability to withstand political manipulation and make impartial decisions during elections. Nigeria provides an excellent example of this popular tendency. During the 2007 elections Nigerians lacked confidence in INEC because the then, Chairman Professor Maurice Iwu, was alleged to be a close affiliate of the President. Popular 206 perceptions of INEC impartiality changed for the better in 2011, when Professor Attahiru Jega, a former university vice-chancellor with a long history of promoting civic rights was appointed as chair (ICG 2011). On election day, citizens are also very cognizant of the perceived impartiality of polling day staff, with the expectation that workers who might be registered members of political parties have greater incentive to bias the results in favor of their party. During the 2004 and 2009 elections in South Africa, supporters of opposition parties criticized the Independent Electoral Commission (IEC) for employing teachers as poll day workers, based on the belief that these workers, many of whom were members of the COSATU trade union — a main affiliate of the ruling ANC— would work to ensure the victory of the ANC. Other important attributes of EMBs that citizens may associate with autonomy include EMBs impartiality in decision-making, public accountability, transparency, and financial independence (Kadima and Booysen 2009). 6.3.5 EMB Capacity While ensuring EMB impartiality is important, citizens also want an election process that is as error free as possible. Ideally, citizens would want to register to vote without any difficulties. They also want to spend the least amount of time casting their ballots so that they return to their daily routines confident that votes are counted accurately and announced in a reasonable time. As a result, EMBs ability to overcome the myriad challenges of election administration is vital to voter confidence. Instead of outlining an expansive list of attributes related to EMB capacity, I will highlight a few that are most likely to shape how citizens view elections. These capacity-related attributes correspond to the most crucial stages of the electoral process: voter registration, voting operations, and the counting and tabulation of the results. Take for example EMBs capacity to 207 conduct voting operations. Citizens may focus on the timely opening of polling stations, the competence and efficiency of the polling workers, and the availability and effective functioning of voting equipment. Lapses in one or multiple dimensions of voting operations can spell disaster for the EMB. For instance, during the 21 of April, 2007 Nigerian elections, polling stations did not open until late in the afternoon in several states including Kaduna and Lagos, 29 while in South Africa during the 2009 national elections, voters in townships in Johannesburg and Cape Town had to stand in line for countless hours, before getting the opportunity to cast their ballots. While lapses in EMB capacity have a negative influence on citizens’ election quality perceptions, the weight placed on these lapses vary by the type of capacity deficit. Citizens’ have shown the ability to adjudge the capacity of the EMB in the context of their country’s level of economic and infrastructural development as well as the EMBs experience with organizing multiparty elections. For instance, citizens in South Africa’s first multi-party elections in 1994 were willing to stand in line for days to cast their ballots. This was not the case during 2009 when polling stations in some urban areas ran out of necessary polling material. Similarly, Nigerians personally expressed their willingness to accept the postponement of the April 2, 2011 National Assembly elections that were delayed due to the late delivery of forms to record the election results, once they were confident that the delays were not politically motivated. From an analytical perspective, I expect citizens’ perceptions of EMB autonomy to have the greatest influence on the idea that the electoral playing field is level and perceptions of EMB capacity to be associated with the incidence of electoral irregularities. In practice, however, autonomy and capacity are interrelated and both may jointly affect the incidence of fraud and 29 See EU EOM (2007) 208 administrative irregularities. Moreover, citizens might still find it difficult to disentangle the two EMB characteristics. 6.3.6 How do Citizens Evaluate EMB Autonomy and Capacity? In addition to outlining some of the main characteristics of EMBs and how they may influence citizens’ election quality perceptions, it is important to understand how citizens get information on EMB autonomy and capacity. First, do citizens rely on direct experience with the EMB or information gathered indirectly through other sources? Second, do citizens focus on attributes related to EMB design or the actual performance? Citizens’ evaluations of EMB autonomy can be based on personal experience with electoral management bodies during different stages of the electoral process. However, because election administration is so complex (both with respect to variation in performance across different regions of a country as well as the intricacies of election operations) for citizens to have a more comprehensive assessment of EMBs they also have to rely on information provided by political parties, civil society leaders, the media and other citizens. When EMB evaluations are formed through direct experience, they are more reliable, but limited in scope. For instance, a voter may assess the impartiality of EMB staff working on election day when she votes, but may not have first-hand knowledge of the impartiality of EMB staff working in other regions of the country. However even at this level, citizens’ evaluations can be affected by 1) prior performance evaluations of institutions in previous elections; 2) partisan affiliations; 3) social and demographic attributes, including education, cognition and ethnicity. Evaluations based on indirect information about electoral processes have greater scope and allow citizens a more comprehensive assessment of the elections. Yet, generating a clear 209 picture through this channel depends on 1) the availability of information; 2) the accuracy and completeness of information; and, 3) citizens’ willingness to seek out information on elections. Availability and reliability of information on elections is usually a function of a free, independent, vibrant media environment that produces high quality information with broad coverage throughout the country. Within such an environment, opposition parties will have more opportunities to publicize fraud and electoral malpractice, while journalists play an active role as watchdogs and whistleblowers during the election. Finally, indirect channels also depend on citizens’ willingness to seek out information. Following empirical studies emphasizing the effect that greater access to media has on voting behavior (Conroy-Krutz 2009) and political attitudes (Bratton et al., 2005), I contended that citizens with greater access to media, will be more informed about various aspects of the elections and thereby more likely to develop accurate assessments of the quality of elections. In assessing EMB autonomy and capacity citizens may also focus on the design or performance of the institution. While, I contend that both design and performance are important sources of information, I expect Africans to rely more on EMB performance and use it as the primary means of developing their election quality opinions. First, EMB performance represents a less expensive source when compared to EMB design. Citizens have multiple opportunities throughout the course of the election period to gain knowledge about EMB performance. This can be done by participating in key stages of the elections such as voter registration and electionday polling; by listening to radio programs or by attending campaign rallies. Compare this to relying on EMB design, which requires in depth knowledge and exposure to legal framework of the EMB. Accessing this form of information may be more costly for regular citizens, and limited to those who are well educated and those interested in politics. It also depends on how 210 issues related to design are publicized in the media or by political parties. In countries under going electoral reform issues related to EMB design are discussed in the public domain (for instance in countries such as Kenya, Zimbabwe and Nigeria). But in other countries, such as South Africa, there is little discourse about EMB design. Second, EMB design is static and represents a snapshot of how the organization is expected to perform according to the rules. EMB performance, on the other hand, is dynamic and reflects factors within and outside of the EMB that affect the functioning of the organization. Nonetheless, EMB design can be very instructive for citizens during the early stages of the electoral process as it signals EMBs’ ability to deter electoral fraud and minimize electoral irregularities. 6.3.7 Observable Implications and Hypotheses The overall argument provides some expectations regarding the effect EMBs on citizens’ perceptions of election quality. First, I expect citizens’ perceptions of election quality to be positively associated with the autonomy and capacity of EMBs, after accounting for other factors associated with the electoral environment and citizens’ individual attributes. • (H1) When EMBs perform well citizens are more likely to consider elections free and fair 6.4 Descriptive Statistics 6.4.1 Africans’ Evaluations of Election Quality in Africa I start by exploring basic trends in Africans’ views towards election quality and election administration in Africa. I examine at citizens’ evaluations of election quality in Africa using data from the Afrobarometer public opinion research project that includes four rounds of surveys 211 from 1999 to 2009. 30 The Afrobarometer survey includes one question that gauges citizens’ opinions of election quality. Specifically, citizens were asked “On the whole, how would you rate the freeness and fairness of the last national election held in XXXX?” The response scale includes “Not free and fair”, “Free and fair with major problems”, “Free and fair with minor problems” and “Completely free and fair”. Across the sample of 20 countries surveyed by the Afrobarometer between 1999 and 2009, 31 a plurality of citizens (46 percent) rated elections as completely free and fair, while one in four citizens rated the elections as free and fair with minor problems. Only 14 percent of respondents had a completely negative assessment of the democratic quality of elections in their countries and an equal proportion (14 percent) identified the elections as free and fair with major problems. In Table 6.1, I disaggregate election quality assessments by three relevant rounds of the Afrobarometer (Round 1: 12 countries; Round 3: 18 countries; Round 4: 20 countries). 32 The round-by-round comparisons allow us to examine the potential changes in the Africans’ election 30 In Round 1, Afrobarometer surveys were conducted in 12 countries from July 1999 through June 2001. Round 2 surveys were carried out from May 2002 through October 2003 in 16 countries (However, the survey in Zimbabwe was completed April and May 2004). Round 3 surveys were conducted in 18 countries from March 2005 through February 2006. Finally, Round 4 surveys were done in 20 countries between March 2008 and June 2009. 31 In Round 1, there were two questions on election quality that differed by country: 1) “In your opinion, were the following elections conducted honestly or dishonestly?” Answers ranged from dishonestly to very honestly; and, 2) “On the whole, how would you rate the freeness and fairness of the last national election?” Answers ranged from completely free and fair to not free and fair. The first question was asked in five countries Ghana, Mali, Tanzania, Nigeria and Uganda while the second was asked in the remaining 7 countries. While the wording for the questions and response options of the two election quality questions are different, they are considered functionally equivalent, especially because they have the same scale. (See Bratton et al. 2005 for the use of the election quality measure based on different questions across a 12country sample in Round 1). In Rounds 3 and 4 of the AB, all countries used the same question: “On the whole, how would you rate the freeness and fairness of the last national election?” Round 2 of the survey did not include questions about election quality. 32 All figures and tables are placed in the Appendix at the end of each chapter. 212 quality perceptions over time. For ease of interpretation, I combine “completely free and fair” and “free and fair with minor problems” into one category indicating high election quality ratings. I also combine “free or fair with major problems” or “not free and fair” to indicate low election quality ratings. Table 6.1 demonstrates a consistent decrease in percentage of citizens who rate elections favorably, falling from a 75 percent circa 1999 (Round 1) to 73 percent circa 2005 (Round 3) and 70 percent circa 2008 (Round 4). While the changes between 1999 and 2005 fall within the margin of error, there is a statistically significant decline between citizens’ election 33 quality evaluations in 1999 and those in 2009. 6.4.2 Africans’ Evaluations of Election Quality by Country A closer examination of election quality opinions for individual countries across the three Afrobarometer rounds may provide further insight. Table 6.2 breaks down citizens who re-ported that elections were either “completely free and fair” and “free and fair with minor problems” for each country included in the Afrobarometer. I find that between 1999 and 2009 there are five countries where a minority of residents has rated elections as either “free or fair with minor problems” or “completely free and fair”. These countries can be categorized as those with the lowest election quality across the sample and include Kenya (2008), Malawi (2005), Nigeria (2005), Zambia (2005; 2009) and Zimbabwe (1999; 2005; 2008). In a second group of countries over 25 percent of citizens (less than 50%) were critical of the quality of elections such as Ghana, Lesotho and Malawi (circa 1999); Cape Verde, Mali and Uganda (circa 2005); and, Lesotho, 33 I also carried out a more precise comparison across the 12 countries that we have data for all three rounds. These include Botswana, Ghana, Lesotho, Malawi, Mali, Namibia, Nigeria, South Africa, Tanzania, Uganda and Zimbabwe. The results are slightly different, but consistent with the overall trend. In Round 1, 75 percent of citizens rated elections favorably compared to 69 percent in Round 2; but in Round 4 there was a marginal improvement to 71 percent. In sum, for these 12 countries the democratic quality of elections has decreased between 1999 and 2008. 213 Liberia, Mali, Senegal, and Uganda (circa 2008). In Botswana, Namibia and Tanzania citizens consistently consider the elections free and fair across the time period. However, in Namibia and Tanzania the perceptions of election quality may be artificially inflated as both countries possess a high proportion of uncritical citizens (Mattes and Shenga 2013). 6.4.3 Africans’ Evaluations of Election Quality Compared to Other Regions of the World How does the quality of elections in Africa compare with other regions of the world? Using the Global Barometer Survey (GBS) data that includes surveys from 55 countries across five world regions (East Asia, South Asia, Latin America, sub-Saharan Africa and the Arab Middle East), we can get an understanding how Africans’ perceptions of election quality compare with citizens living in other regions. 34 Table 6.3 provides details on citizens’ perceptions of election quality across four regions: East Asia, Africa, South Asia and the Arab Middle East. 35 In all four regions, clear majorities of citizens report that they experienced high quality elections. This ranges from a low 61 percent of citizens in the Arab Middle East to a high of 75 percent in East Asia. Africa ranks third among the four regions with 71 percent of citizens report having experienced high quality elections. 6.4.4 Africans’ Trust in EMBs 34 The Global Barometer Surveys was completed in 2006 and includes data from each individual survey region starting from of Round 1 of the East Asia Barometer in 2001 to Round 3 of the Afrobarometer. 35 Responses have been categorized similar to the previous tables where I combine “completely free and fair” and “free and fair with minor problems” into one category indicating high quality elections and a second category “free or fair with major problems” or “not free and fair” indicating low quality elections. 214 Now, I examine Africans’ views towards their electoral management bodies (EMB). Across all four rounds of the Afrobarometer survey, citizens were asked to indicate the level of trust for the national electoral commission in their countries. Between 1999 and 2009, a slim majority of Africans (54 percent) reported trusting electoral commissions either “a lot” or “somewhat”. When I compare citizens’ trust in electoral commissions with other national institutions, I find that Africans support electoral commissions less than their president (63 percent) and courts (58 percent). However, citizens find the police (50 percent) less trustworthy and extend the same level of trust towards the parliament (54 percent). By disaggregating the data on trust across the four Afrobarometer rounds, I see that the proportion of citizens who trust the commission “a lot” or “somewhat” fell from 64 percent in 1999 (Round 1) to 54 percent in 2008-9 (Round 4). This significant reduction in trust over the course of a decade may indicate that electoral commissions have failed to effectively and impartially conduct elections in Africa. 36 Electoral commissions in countries such as Ghana and Tanzania have consistently garnered high levels of trust from over two-thirds of their citizens while electoral commissions in Nigeria (2005- 2009) and Zimbabwe (2005) have not been able to secure the confidence of more than one-third of their populations. From a global perspective, Africans seem less trusting of electoral commissions than their counterparts in East and South Asia. According to the GBS data 42 percent of Africans expressed distrust in their electoral commissions compared to 26 percent in South Asia and 38 percent in East Asia. 36 I also carried out a more precise comparison across the 12 countries that we have data for across all four rounds. The results are slightly different, but consistent with the overall trend. In Round 1, 64 percent of citizens who trust electoral commission “a lot” or “somewhat” compared to 39 percent in Round 2; but in Rounds 3 and 4 there was a significant improvement to 58 and 56 percent respectively. In sum, for these 12 countries citizens’ trust in electoral commissions decreased between 1999 and 2008. 215 6.4.5 Africans’ Awareness of EMBs An important question when considering citizens’ views towards EMBs is whether they have sufficient experience with, or information on EMBs to provide a critical assessment of the institutions’ performance. Building upon the work of Bratton et al (2005: 230), I suggest that Africans rely on factual information when making their assessments of the electoral commission. Moreover, Africans’ awareness of electoral commissions and their democratic role have been increasing over time. Through a close examination of the data, I find that citizens’ trust in the electoral commission is consistent with the actual performance of commissions in recent elections. Kenya provides a perfect example of this. During the 2007—where it was alleged that officials in the Electoral Commission of Kenya (ECK) manipulated election results in favor of the incumbent—we find that citizens’ level of trust fell precipitously from 60 percent in 2005 to 25 percent in 2008. By contrast, Ghana’s electoral commission continues to enjoy broad based trust from approximately two in three Ghanaians (66 percent in 2008) mainly because of the institutions’ high degree of professionalism and impartiality during successive elections (Debrah 2011). The Afrobarometer question on trust in the electoral commission can provide some insight into the extent to which citizens might be informed about the institution. Respondents have the option of reporting, “Don’t know/have not heard enough” in response to the question. I use the percentage of citizens who report not knowing about the institution as an indication of level of (un)awareness of EMBs within a specific country. Africans seems to be least informed about electoral commissions (12 percent) relative to other state institutions such as the president 216 37 (3 percent), parliament (6 percent), courts (6 percent) and police (3 percent). More importantly, however, Africans’ awareness of electoral commissions has shown some improvements over the ten years between 1999 and 2008. The proportion of respondents indicating that they have not heard enough about the institution has experienced a steady decline from 13 percent in 1999 to 9 percent in 2008. 38 For instance, the rate of uniformed citizens fell by 28 percent in Lesotho (between 1999 and 2008) and 11 percent in Senegal (between 2005 and 2008). There are a few countries where the number of citizens who report not knowing enough about the electoral commission has remained high even in the latest Afrobarometer survey such as Madagascar (27 percent), Senegal (19 percent) and Burkina Faso (16 percent). In some countries, citizens have been asked an even more direct question that gauges their knowledge of the electoral commission. Take for example a recent survey conducted in Zimbabwe in May 2012. Citizens were asked whether they had heard about the Zimbabwe Electoral Commission. Only 46 percent of respondents surveyed indicate knowing about the electoral commission. Not only is it important to know whether citizens are cognizant of the performance of election-related institutions such as electoral management bodies, but also whether citizens have adequate information about the design and associated processes of legal and constitutional 37 Afrobarometer gives respondents the opportunity to indicate whether they are not sufficiently informed about particular institutions through the response option “don’t know haven’t heard enough” for all the questions in institutional trust. 38 I also carried out a more precise comparison across the 12 countries that we have data for across all four rounds. The results show a reduction of approximately 50 percent in the proportion of citizens who reported being uninformed about the electoral commission. Specifically, this was witnessed in a reduction of 16 percent in Round 1 to 7 percent in Round 4. One implication of this finding is that within countries with higher more experience with democratic elections, citizens are more capable of developing critically evaluations of electoral commissions. 217 reform of these institutions. Are citizens aware of the variations in the design of electoral commissions? Even more, can we expect citizens to understand how the design of EMBs may influence institutional performance? Unfortunately, there are no standard questions across the Afrobarometer surveys that evaluate citizens’ attitudes towards the design of electoral commissions. However, there are questions from Afrobarometer and other surveys conducted in specific countries such as Nigeria and Zimbabwe that highlight citizens’ general ability to understand complex political information as well as their ability understand issues related to constitutional and legal reforms. In one survey Nigerians were questioned about their awareness of the electoral act passed by the National Assembly before the 2007 elections. This act among many things increased the financial autonomy of the electoral commission. Only 22 percent of citizens reported being aware of the Act. 39 Three years after, Nigerians were also asked a slightly different question about their awareness of the constitutional reform process. Most of the proposed amendments sought to reform election administration by increasing INEC autonomy and capacity. Contrary to 2007, 72 percent of the Nigerians interviewed in 2010 indicated their awareness of the constitutional review process. 40 One reason for significant increase in awareness is that by 2010 Nigerians throughout the country were exposed to information about the reform process through the media and state-by-state stakeholder meetings. In another survey conducted in Zimbabwe in 2012, Zimbabweans were asked about their awareness of the constitution and the ongoing 39 The survey was conducted in February 2007 by the International Foundation of Electoral Studies. 40 The opinion poll was conducted by NOI polls between the 8th and the 14th of March, 2010. The survey involved interviews with 2,058 randomly selected phone-owning Nigerians, aged 18 years and above, across the six geopolitical zones in the country. The telephone numbers were drawn from a database which had been carefully selected and scientifically built to adequately represent all parts of Nigeria. Margin of sampling error of plus or minus 2%. 218 constitution making process. Again, the constitution overhaul spoke to improvements in the election machinery and the quality of elections. Less than one in two Zimbabweans (47 percent) surveyed reported being cognizant of the Zimbabwean constitution while exactly 50 percent were aware the ongoing constitutional process. 41 In brief, this cursory examination of Africans views towards election administration indicates that, with some country exceptions and allowances made for variances in survey questions over time, citizens are becoming more aware of the role of EMBs. In the next section I consider whether EMB performance and design can explain variations in citizens’ election quality evaluations using a statistical analysis. 6.5 Consequences of EMBs on Citizens’ Election Quality Evaluations: Statistical Analysis 6.5.1 Dependent Variable: Citizens’ Election Quality Evaluations The main dependent variable, perceived election quality, captures respondents’ perceptions of the quality of the most recent elections based on a question from the Afrobarometer survey. The main measure of citizens’ election quality perceptions is identical to the one utilized in Chapter 4 (Section 4.4). 42 6.5.2 Independent Variables: EMB Performance 41 The survey was conducted by Freedom House and citizens were asked if they had ever heard about the constitution of Zimbabwe or the constitution making process. 42 Respondents were asked “On the whole, how would you rate the freeness and fairness of the last national election held in XXXX?” Potential responses include “not free and fair”, “free and fair with major problems”, “free and fair with minor problems” and “completely free and fair”. I re-scale the original measure into a dichotomous indicator with 1 indicating that citizens perceived the elections as “completely free and fair” or “free and fair with minor problems”, while 0 denotes the elections as “free or fair with major problems” or “not free and fair.” 219 Of principal concern in this chapter is whether citizens’ evaluations of election quality (dependent variable) are influenced by EMB performance (independent variable) in Africa. The first measure of EMB autonomy and capacity is the EMB performance index obtained from Kelley’s Quality of Elections Database (which is a proxy of the de facto autonomy and capacity of EMBs). This is the same variable used in Chapter 5 based on a additive index of two variables that reflect various attributes of EMB de facto autonomy and capacity during the pre-election and election-day stages of the electoral process (EMB performance index). Across the sample of 28 election periods, the index ranges from minimum of 0.33 in Zimbabwe (1996) and 1 in South Africa (1999 and 2004). 43 The EMB performance index has a mean of 0.73 and a standard deviation of 0.17. As highlighted in the theoretical model, the actual performance of EMBs (de facto autonomy and capacity) may have greater influence on citizens’ election quality perceptions, when compared to EMB design. A cursory examination of the data can provide some initial insights about the relationship between citizen’s election quality attitudes (dependent variable) and EMB performance index (de facto autonomy and capacity). Figure 6.3 displays average scores for the EMB performance index across elections that are either “free and fair” or “not free and fair”. First, EMBs seem to perform (higher levels of de facto autonomy and capacity) better in elections that are free and fair (0.85) compared to those that are not free and fair (0.77). 6.5.3 Election-level and Individual-Level Controls In addition to the attributes of EMB performance, there are certain characteristics of the election period that the existing literature associates with citizens’ opinions of election quality. Because these are the same controls used in Chapter 4 (Section 4), I simply list them: 43 Coding for these variables are described in the Appendix. 220 proportionality of the electoral system and election violence. 44 Additionally, I also include a series of individual-level attributes that the literature highlights as being correlated with perceptions of election quality. These include, partisanship (winning-partisans, losing-partisans and non-partisans), electoral participation (voted in last elections), index of political interests, index of media exposure, education and urban/rural residency. 6.5.4 Multiple Regression Results Do characteristics of EMB performance matter for citizens’ perceptions of election quality in Africa? As you may recall, I argued that attributes of EMB de facto autonomy and capacity (Hypothesis 1) should positively enhance citizens’ election quality perceptions, holding all other covariates constant. The results of the regression are displayed in Table 6.6. 45 Additionally, Table 6.7 provides a more substantive interpretation of the statistically significant predictors by displaying the predicted probability calculations. 46 In Model 1, I examine the effect of EMB performance index (operationalized through an index of the de facto autonomy and capacity) on citizens’ perceptions of the freeness and fairness of elections. The results show that EMB performance has a positive and significant coefficient. 44 I also conduct additional robustness checks by examining other controls highlighted in the literature. These include, level of democracy, logged GNP per capita, time since the last elections, corruption perceptions index and electoral fraud. 45 The table reports the maximum likelihood coefficients (represented in log-odds) of citizens’ likelihood to consider elections free and fair given a unit change in the main individuallevel and election-period predictors. 46 I calculate the marginal change in predicted probability of a respondent perceiving election quality to be free and fair given a minimum to maximum change of a specific independent variable, while holding all non-binary independent variables are held at their mean, while Losing Partisan and Non-partisan are both set to 0, Gender is set at Female, Residential Location is set to Urban, and Proportional Electoral Systems is set to Majoritarian System. Starting from the base model based on the aforementioned profile the average citizens is 49 percent likely to consider the last elections free and fair. 221 The substantive results indicate that Africans in election periods where EMBs display high levels of EMB performance (EMB performance index = 1) are 30 percent more likely to consider election processes free and fair compared to election periods with low levels of EMB de facto autonomy and capacity (EMB performance index = .33). This result provides crucial support for the main hypotheses (H1) by showing that the actual performance of EMBs has a positive influence on citizens’ views on the quality of elections. In fact, when we compare the substantive effect of EMB performance (de facto autonomy and capacity), we see that it has the greatest effect on citizens election quality perceptions based on the cases included in our sample. This effect is more clearly displayed in Figure 6.4 that highlights the change in the predicted probability of citizens free and fair election quality judgments as de facto autonomy and capacity increases. Moving to our control variables. Starting with the proportionality of the electoral system, the estimated coefficient is positive, but its effect on citizens’ election quality perceptions is not statistically significant. I also accounted for the effect of electoral violence during the election period. As anticipated, the coefficient for violence is negative, indicating that citizens have less confidence in elections characterized by violence. In fact, when citizens experience high levels of violence during the election period (Election Violence = 2), the probability that they will consider elections free and fair is reduced by 10 percent, relative to election periods without election violence (Election Violence = 0). As a final step, I consider the effect of individual-level factors on citizens’ election quality perceptions. Commensurate with other studies, the model indicates that losing partisans and non-partisans are less likely to consider elections acceptable, when compared to winning partisans once we control for other covariates. When citizens report an affiliation with the 222 winning party they are 43 percent more likely to espouse free and fair election quality judgements. While on the other hand, citizens who report an affiliation with the losing party or those who report no affiliation are 31 and 19 percent, respectively, less likely to consider elections credible. This finding has crucial policy implications because it shows that groups of citizens in Africa are more inclined to criticize election quality because of their partisan affiliations. Additionally, the results reveal that citizens’ electoral participation has important ramifications. Those who reported having voted in the most recent elections are 8 percent more likely to view elections more favorably, compared to non-voters. To assess the potential mediating effect of citizens’ level of political knowledge, I included a composite index of citizens’ media exposure as well as an indicator of citizens’ level of education. Both the coefficients for media exposure and level of education are negative, but only the latter measure is statistically significant. This could be interpreted to mean that educated citizens are more inclined to criticize the quality of elections, due perhaps to their awareness of the principles, norms and laws guiding elections. The negative estimated effect of education is in contrast with the positive estimated effect of political interests. The model indicates that politically interested citizens are more likely to consider elections free and fair, as the literature intimates this positive effect maybe a result of citizens affiliated with the winner being most interested in politics. Finally, the results indicate that urban dwellers are more critical of election quality, while age and gender fail to achieve statistical significance and were subsequently removed from the model. 6.5.5 Robustness Checks and Alternative Model Specifications 223 To verify the robustness of the findings regarding EMB design and performance, I conducted additional regression analyses. First, I included other potentially confounding factors such as level of democracy, logged GNP per capita, time since the last elections, corruption perceptions index and electoral fraud. As shown in Table 6.8 Models 1-7, the inclusion of these controls has no effect the EMB performance indicator. 6.6 Discussion A main concern for institutionalists is whether political institutions matter; that is, can they effectively explain variations in political attitudes and behavior. In this chapter, I extend this popular line of inquiry by applying it electoral management bodies (EMB) to examine if EMB performance (de facto autonomy and capacity) matters for citizens’ perceptions of election quality in Africa. This chapters’ most important finding relates to the positive and significant effect of EMB performance. The relevance of this finding cannot be overstated, firstly because it is consistent with the main arguments proposed in the chapter that citizens’ election quality perceptions would be most likely be shaped by the de facto autonomy and capacity of the EMB. The election process provides numerous opportunities for citizens’ to directly experience election administration, and the nature of this experience should influence how they evaluate EMBs ability to overcome the challenges of election organization and to ensure that political actors play by the rules. Moreover, citizens’ election quality evaluations may be indirectly influenced by the extent that EMBs constrain the actions of the incumbent and encourage the participation of political opposition. Another reason for highlighting the importance of EMB performance on citizens’ election quality perceptions is because it points to a general trend within the dissertation whereby 224 EMB performance has a positive association with elites’, experts, and citizens’ opinions of election quality. Finally, and perhaps the most important reason for focusing on EMB performance, is that this study, is the first to my knowledge, to consider the effect of EMB performance on citizens’ election quality perceptions using cross-national data. In doing so, the research goes a long way in corroborating existing qualitative and single-country statistical analysis in Africa (as well as other countries) that promotes EMB as institutions with the ability to level the electoral playing field by ensuring that political elite and citizens adhere to the rules of electoral competition. It provides more robust support for explaining why EMBs in South Africa and Ghana have been able to achieve popular legitimacy, while others such as Kenya and Nigeria have not. One main limitation of the analysis involving EMB performance is the inability to desegregate the measure into indicators of de facto autonomy and capacity to examine their relative influence. 6.7 Conclusion This chapter examines whether or not characteristics of EMB performance influence citizens’ election quality perceptions in Africa. Popular media has provided numerous examples that highlight how Africans’ are becoming increasingly aware of the democratic dividends that can be achieved when EMBs are independent of political control and possess the requisite resources to effectively organize elections. Despite, this growing trend, scholars have not done enough to explore the ways that electoral commissions shape citizens election experiences. I argue that EMBs may influence citizens’ election quality evaluations because of the important role the institution plays in reducing the incidence of electoral fraud and minimizing election administrative irregularities. The findings from the statistical analyses provide strong support for the main argument. Most importantly, I find that EMB de facto autonomy and 225 capacity is positively associated with citizens’ views on elections in Africa. Overall the findings point to the relatively important influence EMB performance on citizens’ election quality perceptions. One main limitation of the analysis involving EMB performance is the inability to disaggregate the measure into indicators of de facto autonomy and capacity to examine their relative influence on citizens’ attitudes. In the following chapter, I address this issue in the context of 2007 Nigerian elections through a series of pre and post election surveys that examine citizens’ evaluations of different dimensions of EMB de facto capacity and autonomy and the consequences on perceptions of election quality. 226 APPENDIX 227 Table 6.1: Africans’ Evaluation of Election Quality, 1999-2008 (Complete Afrobarometer Sample) Election Quality Evaluations Round 1 Round 3 Round 4 Total (Circa (Circa (Circa 1999) 2005) 2008) Not free and fair Free and fair, with major problems 25 27 30 28 Free and fair, but with minor problems Completely free and fair 75 73 70 72 Observations 15,952 24,437 27,393 67,673 Countries 12 18 20 50 Note: Question” On the whole, how would you rate the freeness and fairness of the last national election held in XXXX? ” Cell entries represent percentage of respondents. Source: Afrobarometer Rounds 1,3 & 4, with Afrobarometer combined weights. 228 Table 6.2: Africans’ Evaluation of Election Quality Across 20 Countries, 1999-2008 (Complete Afrobarometer Sample) Country Round 1 Round 3 Round 4 Benin – 80 90 Botswana 89 89 94 Burkina Faso – – 79 Cape Verde – 74 82 Ghana 62 82 87 Kenya – 85 20 Lesotho 65 88 64 Liberia – – 73 Madagascar – 86 87 Malawi 65 45 69 Mali 69 71 74 Mozambique – 85 86 Namibia 88 83 86 Nigeria 83 34 33 Senegal – 90 69 South Africa 79 81 78 Tanzania 83 95 92 Uganda 92 72 54 Zambia 79 34 49 Zimbabwe 40 38 48 Note: Percentage of respondents who report that the most recent elections were either “completely free and fair” and “free and fair with minor problems” Source: Afrobarometer Rounds 1,3 & 4, with Afrobarometer combined weights. 229 Table 6.3: Citizens’ Evaluations of Election Quality Across 4 World Regions Region Not Free and Completely Free and Fair Fair Major Minor Problems East Asia 25 75 Africa 29 71 South 28 72 Asia Middle 38 62 East Total 29 71 Note: Cell entries represent percentage of respondents. Source: Global Barometer Survey (GBS) data set with GBS population weights. 230 Table 6.4: Election Periods, Countries and Afrobarometer Rounds Country Benin Botswana Botswana Cape Verde Ghana Ghana Kenya Lesotho Lesotho Madagasca r Malawi Malawi Mali Mali Mozambiq ue Namibia Namibia Nigeria Nigeria S. Africa S. Africa Senegal Tanzania Uganda Uganda Zambia Zambia Zimbabwe Year of Election 2003 1999 2004 2001 1996 2004 2002 1998 2002 2002 1999 2004 1997 2002 2004 1994 2004 1999 2003 1999 2004 2001 2000 1996 2001 1996 2001 1996 Afrobarometer Round 3 Round 1 Round 3 Round 3 Round 1 Round 3 Round 3 Round 1 Round 3 Round 3 Round 1 Round 3 Round 1 Round 3 Round 4 Round 1 Round 3 Round 1 Round 3 Round 1 Round 3 Round 3 Round 1 Round 1 Round 3 Round 1 Round 3 Round 1 231 Table 6.5: Citizens’ Evaluations of Election Quality: 28 Election Periods Response Option Percentag Frequen e cy Not free and fair 11 4,874 Free and fair, with major problems 12 5,214 Free and fair, but with minor 27 6,095 problems free and fair Completely 40 13,366 Don’t Know/ Refused 10 19,827 Total 100 49,376 On the whole, how would you rate the freeness and fairness of the last national election held in XXXX? Cell entries represent percentage of respondents. Source: Afrobarometer Rounds 1,3 & 4, with Afrobarometer combined weights. 232 Table 6.6: Determinants of Citizens’ Election Quality Evaluations Model 1 Election-level EMB Performance (de facto Autonomy/Capacity) 1.594** (0.702) Election Violence -0.545** (0.215) Proportional Electoral System 0.112 (0.277) Individual-level Controls Opposition Partisans -1.326*** (0.035) Non-partisans -0.769*** (0.032) Voted in Last Elections 0.332*** (0.030) Index of Political Interests 0.154*** (0.043) Index of Media Exposure -0.031 (0.057) Level of Education -0.048*** (0.008) Urban Resident -0.108*** (0.030) Random Intercept -1.008*** (0.272) Individual Observations 40,462 Election Periods 28 Note: DV Perceptions of Election Quality (Afrobarometer Round 1, 3 and 4). The analysis is conducted using xtlogit in Stata 10. Cell entries indicate the logistic regression coefficient and the standard errors in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1 233 Table 6.7: Change in Predicted Probability of Free and Fair Election Quality Evaluations Change in Predicted Probability 5% C.I 95% C.I Individual-Level Winning Partisans -0.434 -0.492 -0.35 Losing Partisans -0.31 -0.34 -0.27 Non-partisans -0.19 -0.17 -0.20 Voted in Elections 0.08 0.06 0.09 Interest in Politics Index 0.03 0.02 0.05 Education -0.09 -0.07 -0.12 Urban Resident -0.02 -0.03 -0.01 Election-Level EMB Performance (de facto Auto/Cap) 0.42 0.21 0.60 Election Violence -0.1 -.028 -0.18 Note: All non-binary variables are held at their mean while losers and non-partisans are held at 0,Voters, urban residents and PR held at 1. Predicted probability calculations are based on the Model 3. 234 Table 6.8: Correlates of Citizens’ Election Quality Evaluations (Alternative Model Specifications) Model 1 Election-level EMB Performance (de facto Auto/Cap) Model 2 Model 3 Model 4 Model 5 1.652** 1.628** 1.792*** 1.873*** 2.779*** (0.689) -0.412** (0.204) 0.103 (0.294) – – 0.446 (0.381) (0.666) -0.438** (0.204) 0.259 (0.294) – – – – -0.007 (0.008) (0.716) -0.443** (0.210) 0.198 (0.290) – – (0.849) -0.422** (0.199) 0.321 (0.289) (0.781) Election Violence -0.433** (0.207) Proportional Electoral System 0.152 (0.298) Freedom House Index 0.501 (0.828) GNP per capita (log) Time Since Elections Corruption Perceptions Index -0.007 (0.008) Electoral Fraud 1.065 (0.687) EMB de jure auto/cap Individual-level Controls Opposition Partisans Non-partisans Voted in Last Elections Index of Political Interests Index of Media Exposure Urban Resident Random Intercept Individual Observations Election Periods -1.375*** (0.036) -0.796*** (0.033) 0.336*** (0.031) 0.147*** (0.044) -0.392*** (0.067) -0.097*** (0.030) -1.217*** (0.277) 38,359 27 -1.375*** (0.036) -0.796*** (0.033) 0.336*** (0.031) 0.147*** (0.044) -0.395*** (0.067) -0.097*** (0.030) -1.254*** (0.277) 38,359 27 235 -1.376*** (0.036) -0.796*** (0.033) 0.336*** (0.031) 0.147*** (0.044) -0.392*** (0.067) -0.097*** (0.030) -1.233*** (0.278) 38,359 27 -1.375*** (0.036) -0.796*** (0.033) 0.336*** (0.031) 0.147*** (0.044) -0.392*** (0.067) -0.097*** (0.030) -1.204*** (0.277) 38,359 27 -1.375*** (0.036) -0.796*** (0.033) 0.336*** (0.031) 0.147*** (0.044) -0.394*** (0.067) -0.097*** (0.030) -1.291*** (0.278) 38,359 27 Figure 6.1: Theoretical Model: Consequences of EMB Design and Performance Informal Rules Elite Attitudes & Behavior Formal Rules EMB Design (De Jure Autonomy & Capacity) EMB Performance (De Facto Autonomy & Capacity) Citizen Attitudes Political & Institutional Context 236 Figure 6.2: Citizens’ Evaluations of Election Quality by Election Period Zambia_01 Zimbabwe_96 Nigeria_03 Malawi_04 Lesotho_98 Mali_97 C.Verde_01 Ghana_96 Malawi_99 Mali_02 Zambia_96 Mean Uganda_01 Mozambique_04 S. Africa_99 S. Africa_04 Benin_03 Nigeria_99 Ghana_04 Namibia_94 Madagascar_02 Senegal_01 Lesotho_02 Namibia_04 Uganda_96 Kenya_02 Tanzania_00 Botswana_99 Botswana_04 0.20 0.30 0.40 0.50 0.60 0.70 0.80 Note: Mean Election Quality Perceptions: 1 = “completely free and fair” or“ free and fair with minor problems”; 0 = “free or fair with major problems” or “not free and fair”. Source: Afrobarometer Rounds 1,3 & 4, with Afrobarometer combined weights. 237 Figure 6.3: Variations in EMB Performance 0.85 0.86 0.84 0.82 0.8 0.78 0.77 0.76 0.74 0.72 Not Free and Fair Free and Fair EMB Performance Index 238 Predicted Probability of Free and Fair Election Quality Figure 6.4: Effect of EMB Performance on Citizens’ Election Quality Evaluations 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 EMB Perfotmance Index EMB Performance Index 5% C.I. 239 95% C.I. 0.9 1 CHAPTER 7 THE CONSEQUENCES OF EMB PERFORMANCE ON CITIZENS’ ELECTION QUALITY EVALUATIONS: A CASE STUDY OF NIGERIA (1999-2011) 7.1 Introduction The main finding of the previous chapter is that the performance of electoral management bodies (EMBs), measured through an index of aggregate EMB de facto autonomy and capacity, is positively related with citizens’ perceptions of election quality across 18 Africa countries. This chapter takes a different approach to assessing the link between EMB performance (independent variable) and citizens’ election quality perceptions (dependent variable). Here, I examine citizens’ own evaluations of EMB performance (perceived autonomy and capacity) and gauge whether these evaluations are associated with perceptions of election quality. Popular perceptions of the performance of EMBs are perhaps as important as the actual performance of the institution. This is because perceptions often form the basis of popular democratic attitudes and behavior (Elklit and Reynolds 2005b). Therefore, directly gauging the heterogeneity in citizens’ evaluations of EMB performance will advance our understanding how EMBs influence citizens’ election quality judgements. It is important to understand that elections are, for the most part, localized events that are influenced by a wide range of factors. Because of this, citizens are likely to have dissimilar evaluations of the autonomy and capacity of EMBs, even within a given country during a specific election cycle. For instance, citizens may observe that local election officials in one region may arrive at their polling stations later than officials in other regions. Meanwhile, citizens in one constituency may experience more instances of collusion between EMB officials 240 and the incumbent representative because of intense competition the incumbent faces during the election. Nigeria during the 2007 elections provides a good example of the significant variation in citizens’ EMB evaluations. Citizens were asked how they rated the performance of Nigeria’s main EMB – the Independent National Electoral Commission (INEC). Although international observers and domestic elites described INEC’s performance as the worst since the reintroduction of democratic rule in 1999, there was significant heterogeneity in popular evaluations of INEC performance. Over 35 percent of Nigerians surveyed rated the performance 1 of INEC as fairly good (29 percent) or very good (8 percent). I find similar heterogeneity in EMB performance evaluations in Zimbabwe after the botched 2008 elections, despite consensus among stakeholders that the Zimbabwe Electoral Commission (ZEC) was complicit in delaying the announcement of the results of the presidential elections. The first purpose of this chapter is to explore the association between citizens’ evaluations of EMB performance (the independent variable) and their perceptions of election quality (the dependent variable). Here I build upon the argument outlined in Chapter 6 and demonstrate how citizens’ evaluations of EMB performance has a positive impact on their election quality perceptions. This is because EMBs with high levels of autonomy and capacity are more likely to reduce electoral fraud and minimize electoral administrative irregularities. I empirically assess this relationship using public opinion data from Nigeria during the 2007 election process. The findings underscore the strong and positive association between Nigerians’ evaluations of INEC and their perceptions of the democratic quality of the 2007 elections. In fact, 1 In 2007, the Afrobarometer asked respondents to evaluate the performance of INEC during the last elections. The variable is coded from 0 (very poor) to 3 (very good). In 2007, INEC performance evaluations had a mean of 1.11 and a standard deviation of 0.98. 241 citizens’ evaluations of INEC were the most important predictor of the freeness and fairness of the 2007 elections in Nigeria, even after accounting for performance of other election-related institutions, features of the electoral environment, and citizens’ individual attributes. To put it simply, Nigerians were 38 percent more likely to report that elections were flawed when they believed that INEC performed poorly in the elections. Because there are several characteristics of EMBs that are associated with their performance, the second purpose of the chapter is to examine different dimensions of EMB performance from the perspective of citizens. As in Chapter 5, I distinguish between EMB autonomy and capacity and consider whether citizens are able to distinguish between these two dimensions of EMB performance. Finally, by empirically distinguishing between citizens’ evaluations of INEC autonomy and capacity, I consider whether one of the two dimensions has a stronger association with citizens’ election quality perceptions. Using data on the 2007 Nigerian elections I find that citizens are able to distinguish between the autonomy and capacity of EMB performance. However, it seems that Nigerians’ perceptions of INEC autonomy played a greater role in their overall election quality judgements. I explore these lines of inquiry within the context of Nigeria because the country provides one of the most compelling cases of the significant influence of electoral management on popular attitudes towards elections. From one perspective, the 2003 and 2007 Nigerian elections highlighted the multiple ways in which an ineffective and biased EMB could undermine popular confidence in the quality of elections. From another perspective, Nigeria’s 2011 elections represented a significant departure from previous multiparty elections: the implementation of crucial legal and procedural reforms helped enhance INEC’s preparedness and insulation from political influence. These reforms helped transform the institution from one characterized by 242 public ignominy to one with a groundswell of popular approval. Another reason for focusing on Nigeria is the availability of public opinion data that gauges citizens’ views on election quality as 2 well as their evaluations of EMB performance across different election periods. Finally, because of the important reforms that were implemented ahead of the April 2011 elections, Nigeria represents an ideal case study on the effect of election administrative reforms on citizens’ election quality perceptions. So the third and final purpose of the chapter is to outline possible ways through which citizens’ election quality perceptions and EMB performance evaluations are affected by election administrative reforms. The remaining chapter is organized as follows. Section 7.2 provides a brief history of elections and election management in Nigeria between 1999 and 2011. In Section 7.3, I focus on how citizens evaluate EMB performance and describe the measures of EMB autonomy and EMB capacity, while in Section 7.4 I assess the relationship between the two dimensions of citizens’ EMB performance evaluations and their perceptions of election quality. I conclude with a brief summary of the findings and a discussion of the main theoretical and policy-related implications. 7.2 Elections and Election Management in Nigeria (1999-2011) 7.2.1 Elections in Nigeria 1999-2011 Nigeria’s most recent experiment with democracy began in 1999 and since then the country has experienced three democratic elections in 2003, 2007, and 2011. During the 12 years 3 of uninterrupted multiparty rule, there are have been two main phases in the quality of elections. First, between 1999 and 2007 the quality of elections declined precipitously. Each progressive 2 I also had the opportunity to gather original data in Nigeria on citizens and elites attitudes towards the elections and election management through fieldwork conducted in 2010 and 2011 3 See Table 7.1 for a list of Nigeria’s democratic elections since independence. 243 contest was characterized by even more egregious instances of violence and intimidation, systematic electoral fraud and manipulation, and a patently partial and ineffective election administrative body. The 2011 elections marked a significant departure from the ignominy of past elections and a new phase of democratic elections. Through comprehensive electoral reforms, the appointment of a new chairperson of the Independent National Electoral Commission (INEC), and commitments to credible elections by the then incumbent president, Nigerians experienced the most democratic elections in the country’s history (Lewis 2011; LeVan and Ukata 2012; Aiyede 2012). 4 According to the Nigerian Constitution (1999) and the Electoral Act (2003, 2006, 2010), the Independent National Electoral Commission (INEC) functions as the main electoral management body in Nigeria. INEC is granted powers to organize Federal and State elections, to register voters and political parties, and to monitor party financing and political campaigns (Jinadu 2011). Since the establishment of INEC for the 1998-1999 election period, INEC’s public image has been sullied by a lack of institutional autonomy, political neutrality, professionalism and administrative efficiency (Ibid). Numerous constitutional, statutory, and procedural reforms have been implemented to address the challenges. However, the success of these reforms depended on the political will to enhance the quality of election administration as well as the legitimacy of elections in Nigeria. 7.2.2 1998-1999 Elections: Rigged, but Better than a Return to Military Rule 4 In spite of the reduction in fraud and administrative irregularities, the 2011 set a record with the number of election-related fatalities (800), making it the most violent in the country’s history (HRW 2011). 244 On February 27 1999, Nigerians elected Olusegun Obasanjo of the Peoples Democratic 5 Party (PDP) as President. In addition to receiving 63 percent of the vote share in presidential 6 election, the PDP gained a majority in both houses of the National Assembly. For the most part, international and domestic observers reported that the elections were peaceful and devoid of major acts of election-related violence (Kew 1999; Nwankwo 1999; TCC/NDI 1999). The main problem, however, was the widespread occurrences of electoral fraud, including the falsification of results, stuffing of ballot boxes, and the intimidation of voters by thugs affiliated with the main political parties. Systematic and widespread incidents of electoral fraud were most evident during the presidential elections in the South-South and South East geo-political zones of the 7 country. The main opposition presidential candidate, Olu Falae of the Alliance of Democracy, complained bitterly about the flagrant manipulation of popular will, explaining that the degree of fraud was “so monumental as to make nonsense of the entire process” (Washington Post, March 02, 1999). Moreover, a group of international observers led by former U.S. President Jimmy 5 Nigeria’s founding elections during the “Third Wave” were held on June 12 1993 after the 8-year military rule of Gen. Ibrahim Babangida (Edozie 2002). Chief M.K.O Abiola won the election that was considered by major stakeholders to be free, fair, and peaceful. However, General Babangida derailed the democratization process by annulling the results of the elections and in- stalling a temporary civilian government. The period of instability provided fodder for General Sani Abacha to overthrow the civilian government and embark on a 6-year brutal and repressive military rule that witnessed the abrogation of human rights and civil liberties. Moreover, leaders of Nigeria’s prodemocracy movement were detained and some, including Ken Saro-Wira, were killed. Abacha’s rule came to an abrupt end after he suffered a heart attack on June 8. This unexpected event, along with the unexplained death of Chief Abiola, provided the context for a transition program that commenced with the local government elections in 1998 and the presidential elections in February 1999 (Edozie 2002). 6 See Table 7.2 and Table 7.3 for the results of the presidential and National Assembly Elections, respectively, between 1999 and 2011. 7 For example, Darren Kew, a prominent political scientist and international observer during the 1999 elections in Rivers state, reported witnessing polling officials and party agents thumb printing all the allotted ballot papers so as to ensure 100% turnout for the local district (Kew, 1999). 245 Carter failed to endorse the final results of the presidential election because the level of fraud was so extensive that it prevented the observers from making an “accurate judgment” about the outcome (TCC/NDI 1999). Nonetheless, most stakeholders were willing to accept the imperfect conduct because securing a peaceful transition to civilian rule was more important than the fairness of the elections (Lewis 2003:133 ;Ogbeidi 2010:57). Many stakeholders commended the newly created Independent Electoral Commission 8 (INEC), under the leadership of Justice Ephraim Akpata, for its autonomy during the election 9 cycle (Nwankwo 1999; TCC/NDI 1999). INEC had a limited time to prepare for elections, and this impacted the organizations’ voter registration and election-day capacity. Despite these challenges, however, INEC improved its performance during the election cycle. By the presidential elections on February 27, Nigerians witnessed polling stations that opened on time with sufficient voting materials. While stakeholders were relatively satisfied with INECs performance, they suggested several reforms that were vital to the credibility of upcoming elections. 10 7.2.3 2003 Elections: Massive Fraud and Opposition Boycotts The 2003 elections marked a milestone in the country’s democratic history: the first civilian administered elections in over two decades (LeVan, Pitso and Adebo 2004; Lewis 2003). As the results of the April 19 presidential elections were announced, it was clear to most 8 9 INEC was established in 1998 through Decree N. 17 of 5 August 1988. Nonetheless, Kew mentions that there was general confidence in INEC at the federal and state level but the local poll officials were susceptible to corruption as they were directly involved in the election fraud. (Kew 1999: 30). 10 For instance, The Carter Center/National Democratic Institute recommended making changes to the legal framework to guarantee INEC greater financial autonomy and autonomy in the appointment process of electoral commissioners. 246 observers that the 2003 elections failed to meet basic domestic and international democratic standards. In contrast to the 1999 elections, 2003 experienced even more systematic rigging, manipulation, and fraud (Lewis 2003: 141). According to reports from Transition Monitoring Group (TMG), 11 a coalition of domestic observer organizations with approximately 10,000 observers in states in the South-South and South-East geopolitical zones (such as Edo, Delta, Bayelsa, Cross River, and Anambra) political thugs were seen chasing voters and observers from the polling stations, and stealing and stuffing ballot boxes (TMG 2003; LeVan, Pitso and Adebo 2004). While most domestic and international observers described the actual elections as relatively peaceful and orderly, there were over 100 election related deaths and many more Nigerians were injured between April and May, during the height of the election campaign period (EU EOM 2003; HRW 2011). 12 In the run up to the 2003 elections, the National Assembly passed the Electoral Act (2002) which gave INEC authority for compiling the voter register and mandated for counting to take place at the polling station. Additionally, INEC improved its voter registration capacity by employing a computerized voter registration system and also distributed new voter registration cards (EU EOM 2003; TMG 2003). Unlike 1999, international and domestic observers heavily criticized INEC for its lackluster management of the electoral process (EU EOM 2003; TMG 2003; JDPC 2003). Although INEC instituted a series of reforms to enhance its capacity, including the creation of computerized register, these changes were not enough to offset crucial shortcomings in its 11 The Transition Monitoring Group is a coalition of 170 domestic human rights and civil society organizations in Nigeria. 12 According to a Human Rights Watch Report, members and supporters of the ruling PDP perpetuated most of the violence (HRW 2003). 247 logistic operations. 13 For instance, the voter registration process was delayed and lacked transparency, and on election day there was widespread chaos and confusion as polling stations opened late, lacked adequate voting material, and voters were denied ballot secrecy (NDI 2002; IRI 2003:16; TMG 2003:37). To make matters worse, opposition parties alleged that INEC was actively colluding to rig the elections. To be fair, the European Union Observation Mission, in its final report, noted the challenges that INEC faced in organizing elections such as the size of the electorate (largest in Africa) and infrastructural deficiencies. But the report also pointed to the problem of executive control over the appointment of commission members (at the national and state levels) and the lack of financial autonomy as the principal impediments to popular confidence in INEC (EU EOM 2003). Finally, extensive electoral fraud and administrative problems prompted opposition parties to call for the reconstitution of INEC and a rerun of the Presidential and National Assembly polls. In particular, opposition presidential candidate of the All Nigeria Peoples Party (ANPP), Muhammadu Buhari, threatened to boycott the upcoming state legislative elections unless these conditions were met (Lewis 2003: 143; Ogbeidi 2010: 51). Unfortunately, these demands fell on deaf ears; most opposition parties and their supporters “virtually” boycotted the state elections. In the aftermath of the elections, over 560 elections the Electoral Tribunal. 14 petitions were lodged with 15 7.2.4 2007 Elections: “Worst elections that ever took place in the country” (TMG) 13 14 See Table 7.7 for a list of electoral reforms implemented between 1999 and 2011. See Jinadu 2011. 15 The 1999 Nigerian Constitution (Section 285) and 2006 Electoral Act (Part IX) empowers the Electoral Tribunal to hear petitions brought forward by candidates and political parties. See Table 7.6 for the number of post-election petitions between 1999 and 2011. 248 The 2007 elections represented the third consecutive election since a democratic transition in 1999. It was also the first time that Nigerians would witness civilian turnover in leadership because the incumbent President, Olusegun Obasanjo, was ineligible to run again at the end of a two-year term in office (Ibrahim 2007). Despite the historical significance, most stakeholders regarded the 2007 general elections as the worst in the country’s history: it was a continuation of a downward spiral in election quality evident in 1999 and 2003 elections (Rawlence and Albin-Lackey 2007; Herskovits 2007). During the 2007 election period, Nigerians witnessed widespread electoral irregularities such as underage voting, ballot box stuffing, voter intimidation, and false announcement of results on an unprecedented scale. The magnitude of electoral violence eclipsed that experienced in 2003. There were numerous political assassinations and approximately 280 deaths between November 2006 and April 2007 (ICG 2007; NDI 2007). Local and international observers condemned political parties for perpetuating a culture of violence, political assassinations, and vote rigging. Security forces, mainly the Nigerian Police Force, were blamed for their blatant support of political candidates and their inability to maintain law and order (TMG 2007; EU EOM 2007). As in 2003, electoral administrative reforms were introduced through the 2006 Electoral Act. The act sought to address the problem of INEC’s administrative autonomy by giving the commission full authority in selecting the Secretary of the commission who serves as the chief electoral administrator. The new statute also granted INEC greater financial autonomy through the establishment of a dedicated fund for INEC’s operational expenses. The commission also embarked on a new electronic voter registration system that would reduce fraud and double registration by capturing citizens’ biometric data. 249 Despite these reforms, stakeholders censured INEC for its lack of independence from executive control, low levels of transparency, and a complete lack of administrative preparedness. The voter registration process was marred by technical and logistical failures. Politicians also hijacked the registration process by inflating the voters’ roll with fictitious names such as “Mike Tyson”. Although INEC was able to register approximately 60 million Nigerians, many stakeholders questioned the accuracy and reliability of the voters’ roll and in the end INEC never circulated an official copy of the voter register (TMG 2007). In assessing INEC’s logistical capacity on election day (delivering voting material and opening polling stations), the National Democratic Institute, an international observer group, reported that the scale of unpreparedness was the worst they had observed anywhere in the world. 16 For many election stakeholders, the performance of INEC was so abysmal that it seemed as if INEC was “programmed to fail” by the ruling PDP (TMG 2007). In fact, there were numerous allegations of collusion between INEC officials and political elites. Another source of concern was INEC’s lack of financial independence. Although the Electoral Act (2006) established the INEC Fund to increase the institution’s financial autonomy, this fund was not implemented during the 2007 election period. Instead, various civil society groups criticized President Obasanjo for allegedly using his control over the commission’s purse strings to frustrate the effective and timely execution of key aspects of electoral administration (EU EOM 2007; NDI 2007). The incumbent Peoples Democratic Party (PDP) swept the elections winning the presidency, a super-majority in the National Assembly, and most of the state governorship races. 16 Additionally, NDI in its final report lamented INECs reluctance to accredit voters, illegal disqualification of candidates, its failure to adequately train election-day workers, and the failure to release election results in the presidential election even up to this day. 250 However, the conduct of the elections brought so much international and domestic condemnation that in his inauguration speech the newly elected president, Umaru Yar’Adua, admitted to the illegitimate basis of his election and committed to broad-based electoral reform (Yar’Adua 2007). Another indication of the woeful conduct of the 2007 elections was seen with the number of petitions filed with the Election Tribunals. In 2007, over 1250 election petitions were filed, compared to 560 in 2003 and 400 in 2011 (Jinadu 2011). The nature of the irregularities was so pronounced that tribunals overturned many of the certified election results. For instance, elections results were vacated in 9 out of 36 gubernatorial elections. 7.2.5 2011 Elections: “Substantial improvement” but “far from perfect” 17 On May 4th 2011, President Obama called the newly elected Nigerian President, Goodluck Jonathan, to congratulate him on his victory in the April 16th Presidential election (White House May 4, 2011). 18 During the conversation, Obama lauded the Nigerian people, INEC, and Nigerian civil society for ensuring the success of the 2011 elections. But he also denounced the post-election violence that claimed the lives of 800 Nigerians. For most stakeholders, the 2011 elections represented the most credible poll since the re-introduction of democratic rule in Nigeria in 1999 (EU EOM 2011; Akhaine 2012). 19 While there are many reasons for the improvements in the integrity of the 2011 elections, I focus on the three most important. 17 18 See Clinton (2011). The White House, Statement by President Obama on Elections in Nigeria, May 4, 2011. 19 As indicated in Chapter 2, I had the opportunity to serve as a long-term observer for the National Democratic Institute during the 2011 elections. I was deployed in the South West zone of the country (Ekiti, Lagos, Ondo, Osun, Oyo, Ogun) for 5 months and conducted informal interviews with elites and citizens. Based on my own observations the 2011 elections experienced marked improvements with the overall quality of elections as well as the autonomy and capacity of INEC, relative to what had been reported in 2007. 251 First was the credible commitment of the Nigerian president (Goodluck Jonathan) and other political elite to improve the quality of elections in Nigeria. As early as 8 months before the actual elections, the President publicly announced his commitment to improving the quality of elections in Nigeria. This commitment to free and fair elections went along way in stemming the culture of divisive inter-and intra-party competition, impunity from prosecution for electoral offenses, and abuse of state institutions that typified Nigerian elections (Aiyede 2012:2). 20 It was also a stark departure from the 2007 elections, where the then incumbent president, Olusegun Obasanjo, publicly stated that the election as a “do-or-die affair” (JDPC 2007; Adebayo and Omotola 2007). The second reason for the historic elections was the successful passage of important electoral reforms in 2010 and 2011. 21 These reforms were enshrined in the 2010 amendments to the 1999 Constitution and the 2011 Electoral Act. The reforms sought to enhance the autonomy and capacity of INEC; promote internal democracy of political parties; deter electoral 20 Through my own interaction with elites and citizens during the elections you got the sense that politicians and their followers were guided by the fear of prosecution but also the desire to restore Nigeria’s sullied international image. As a result, the country experienced less incidents of political violence and electoral fraud during the campaign and election periods when compared to the 2007 elections. Even in defeat, many incumbent governors and members of the National Assembly demonstrated a degree of civility and cordiality towards their opponents that was unheard of in previous elections. Moreover, state institutions such as the police and military acted with greater levels of impartiality and professionalism during the elections. It seemed as if all government employees had been given the same talking points: we were com- mitted to free and fair elections, and even if this was a nominal support, it made a difference given Nigeria’s history of fraudulent elections. 21 The electoral reform process was initiated in 2007 by then President Umaru Yar’Adua who established the Electoral Reform Committee (ERC). The ERC was tasked with developing a set of reform proposals that would seek to improve Nigeria’s electoral system. The ERC submitted its final report to the President and the National Assembly in December 2008 that included proposals for addressing the independence of INEC, promoting internal party democracy, and prosecuting electoral offenders. However, it was not until July 2010 that the National Assembly passed the bill to amend the constitution. 252 malpractice; and expand the capacity of Election Petitions Tribunals (EU EOM 2011). As it pertained to election management, one of the most important aspects of the constitutional reform was the strengthening of INEC administrative autonomy vis-a-vis the executive. According to Section 160 of the 1999 Constitution (amended) “[...] in the case of the Independent National Election Commission, its powers to make its own rules or otherwise regulate its own procedure 22 shall not be subject to the approval of the President.” However, one important aspect of INEC autonomy that was not included in the amendments was the appointment process of INEC national and regional commissioners. This still remained the prerogative of the executive. Third, was the appointment of new chair of INEC, Professor Attahiru Jega in June 2010. Professor Jega was a well-known political scientist, member of the Electoral Reform Committee, and a former head of the National Academic Staff Union – a leading trade union in Nigeria. Professor Jega’s reputation as an impartial, hardworking professional, with long track record 23 within civil society instilled widespread popular confidence in the commission. For the first time since 1999, Nigerians were convinced that the electoral commissioner was not beholden to the President. A leader of a prominent civil society organization in Lagos explained, We are confident that the appointment of Jega is a game changer. There is no way that INEC will achieve 100 percent confidence from the public and major stakeholders, but you will have 60 percent credibility in the electoral system. Change in the governance will increase 60 percent of the responsiveness. 22 23 See Table 7.7 for a detailed list of the reforms introduced in 2010 and 2011. According to a leading opposition member in the Senate, “the president did a good job in selecting Jega who is independent and credible. For the president to appoint Jega it indicates a will to bring about electoral reform. He is certainly taking a political risk of appointing someone like Jega” (Abuja, June 2010). 253 In his capacity as INEC Chairman, Jega worked to internally reorganize the commission by developing clear mechanisms of accountability and threatening to remove those officials engaged in corruption. 24 He also cultivated strong linkages with civil society and political parties and he was always willing to consult election stakeholders when the commission faced challenges (ICG 2011). Fourth, Jega and the commission established a series of procedural reforms aimed at improving transparency and administrative efficiency of INEC; see Table 7.7 for a detailed description of these changes. Some of the most meaningful reforms included the creation of new biometric voter’s register, the introduction of the modified open-ballot voting system, the use of NYSC members as temporary poll workers, and posting results at polling stations (NDI 2012). These reforms helped to restore a significant proportion of the confidence that had been lost since in 2003 and 2007 elections. INEC received widespread acclaim from international and domestic stakeholders for its improved management of the elections. 25 As with any election exercise, INEC faced a series of logistical challenges. For instance, with the rollout of the voter registration system in January 2011 INEC staff experienced technological glitches with the datacapturing machines, but these difficulties were resolved before the completion of the voter registration. As a result, for first time INEC was able to conduct a registration process and 24 Through informal conversations with INEC staff, directors, and Regional commissioners, many expressed a commitment to free and fair elections, but also feared being removed from their position either because of they were perceived as being biased towards a particular party or because of poor performance. 25 While citizens were reluctant to immediately invest trust in the institution, during the course of the elections popular support for the commission swelled. As Akhaine (2011) points out, and I was able to experience for myself, while serving as an international observer, citizens went above and beyond their obligations to ensure the success of the commission. On one occasion, I witnessed private citizens donating printer cartridges so that the local election officials could effectively print voter ID cards in a small community in Oyo State. 254 produce a voter register, with over 70 million voters, that most stakeholders regarded as legitimate. Despite the improvements in election administration and political will, the 2011 elections were far from perfect (Obama 2011). In some regions of the country, especially the South- South and South East, observers reported instances of electoral irregularities. Generally, however, Nigerians never experienced the level of fraud and intimidation that plagued the 2007 elections (EU EOM 2011). On the other hand, the levels of violence during the elections exceeded all the other elections combined (Lewis 2012). For most observers, the violence was linked to Northern reaction to the election of Goodluck Jonathan, a Christian from the south of the country who was perceived to have robbed the North of its opportunity to lead the country and to violate an informal understanding of “zoning” whereby presidential terms would rotate between individuals from the Northern and Southern regions of the country (Ibid). 26 In the following section, I examine Nigerians’ attitudes towards INEC and explore their evaluations of INEC autonomy and capacity using recent public opinion data. 7.3 Nigerians’ Evaluations of the Independent National Electoral Commission (INEC) 7.3.1 Nigerians’ Trust in INEC One way to gauge Nigerians views towards INEC since 1999 is to assess the level of trust they have in the institution. 27 The Afrobarometer survey asked Nigerians how much they trust 26 Additionally, the outbreak of violence may have been directly linked to the perceived manipulation of the vote in many geo-political zones of the South South and South East. For instance, in some areas over 100% of registered voters supported Good Luck Jonathan (Lewis 2012). 27 The focus on institutional trust is informed both by the existing literature and fieldwork experience. Following the work by Mishler and Rose (2001), institutional trust can reflect both citizens’ most recent evaluation of the institution’s performance as well as historical evaluations. 255 INEC following the four elections since the reintroduction of multiparty rule in 1999. 28 As shown in Figure 7.1, the number of citizens that trust INEC “a lot” or “somewhat” (indicating trust) has fluctuated over time. 29 The highest levels of trust were recorded after the 1999 elections, where 6 out of every 10 Nigerians found the newly minted institution trustworthy. Since then, popular trust in INEC plummeted precipitously to 10 percent of Nigerians following the 2003 elections. Meanwhile, there have been marginal improvements in popular trust in INEC after the 2007 elections (28 percent) as well as the 2011 elections (33 percent). In many respects, the overall trend in popular trust in INEC mirrors the deterioration in the institutions’ performance since the 1999 elections. However, the results following the 2011 elections confirm that popular trust in INEC is rising (relative to evaluations of INEC performance). Although INEC’s performance improved significantly during the 2011 elections (as will be documented below), citizens seem cautious about investing confidence in the institution. It may take a series of elections cycles in which INEC performs effectively for citizens to begin to express greater levels of trust in the institution. 7.3.2 Nigerians’ Evaluations of INEC Performance Citizens’ attitudes toward INEC are also reflected in their evaluations of the institutions’ performance. Following the 2007 and 2011 elections, Afrobarometer asked Nigerians to assess the performance of INEC (See Table 7.9). In 2007, a plurality of respondents rated INEC’s While I expect citizens’ trust in the EMB to be closely related to their performance evaluations, there are instances in which the two attitudes may deviate: citizens might not trust an institution given its long history of producing unfavorable outputs, but yet believe that it currently performed well or above expectations. 28 Afrobarometer surveys were conducted in December 1999 (Round 1), October 2003 (Round 2), August 2005 (Round 3), May 2008 (Round 4), and November 2012 (Round 5). 29 All figures and tables are placed in the Appendix at the end of each chapter. 256 performance as “very poor” (33 percent), while less than 1 out of every 10 Nigerians considered the institution to have performed well during the 2007 elections (7 percent). The improvement in INEC performance during 2011 elections cycle is clearly reflected in citizens’ performance evaluations. According to the Afrobarometer survey conducted in 2012, a majority of citizens reported INEC performance as “fairly good” (51 percent) and a further 20 percent gave the institution a rating of “very good”. When these two favorable ratings are combined, over 70 percent of Nigerians were pleased with INEC’s performance. If we compare citizens’ evaluations of INEC between 2007 and 2011, we find that 34 percent more Nigerians came to view INEC performance positively. 7.3.3 Nigerians’ Evaluation of INEC Autonomy and Capacity In addition to gauging citizens’ overall evaluations of INEC performance, I also distinguish between Nigerians’ evaluations of INEC autonomy and INEC capacity. As one may recall, distinguishing between autonomy and capacity of EMBs is important because it expands our understanding of the mechanisms through which EMBs influence citizens’ attitudes and behaviors. To examine Nigerians’ evaluations of INEC autonomy and capacity, I utilize data from International Foundation of Electoral Systems (IFES) post-election survey conducted in Nigeria immediately after the 2007 elections. 30 7.3.4 Evaluations of INEC Autonomy (2007 Elections) Again, I define autonomy as an institution’s ability to make decisions independent of the control of the executive and political party elite, as well as other state and society actors. EMB 30 IFES conducted the survey between May 3 and 10 with a nationally representative sample of 2,416 Nigerians that allows for national inferences based on ±2 margin of error at a 95% confidence level. 257 autonomy is important for citizens because it reflects the extent to which the incumbent and other political elites can manipulate the EMB and engage in other forms of electoral fraud. IFES asked respondents a number of questions that taps into their evaluations of INEC autonomy during the 2007 elections. First, respondents were asked whether or not “INEC performs its duty as a neutral body guided in its own work only by the law, or INEC makes decisions which favor particular interests?” This question represents a general assessment of INEC autonomy that most effectively captures my conceptualization of EMB autonomy. When responding to this question, citizens may reference information on the impartiality of INEC as a national institution (such as the reputation or background of the EMB chairperson), or more specific information about INEC in their state or local government area. A second more specific measure of autonomy reflects citizens’ evaluations of the impartiality of polling staff on election day. This question taps into citizens’ most direct experience with INEC officials and provides important insight about the perceived level of EMB autonomy within a specific region. The third measure of INEC’s autonomy explores citizens’ perceptions of the impartiality of vote counting and tabulation process. Citizens’ satisfaction with the counting and tabulation of votes represents another important dimension of their assessments of EMB autonomy. Because EMB officials are directly responsible for the vote counting process, the extent to which they are insulated from external influences has important implications for how citizens view the integrity of the entire election process. I construct a composite index of EMB autonomy from the aforementioned three indicators: 1) the overall neutrality of INEC, 2) the impartiality of poll workers, and 3) the 258 impartiality of the counting and tabulation processes. 31 The autonomy index ranges from 0 to 1 with increasing levels of perceived autonomy (mean = 0.49; SD= 0.36). 32 Table 7.10 displays citizens’ evaluations of the three individual indicators of INEC autonomy as well as the autonomy index. When we compare the three indicators of autonomy, citizens are most critical of INEC’s overall neutrality (mean=0.43), followed by the impartiality of counting process (mean= 0.51), and the impartiality of poll workers (mean= 0.58). 7.3.5 Evaluations of INEC Capacity (2007 Elections) The second dimension of EMB performance is institutional capacity. In essence, capacity reflects an institution’s ability to implement its decisions and legally mandated responsibilities in an effective and efficient manner. I gauge perceived capacity through a composite index constructed from four questions on the IFES survey. These include citizens’ satisfaction with (1) the voter registration process, (2) the adequacy of information regarding the poll location and voting procedures, (3) the competence of poll workers, and (4) the adequacy of poll facilities. The capacity index ranges from 0 to 1 with increasing levels of perceived capacity. 33 First, let us consider citizens’ evaluations of the voter registration process. EMBs’ ability to effectively register voters provides citizens’ with the glimpse of the institutions’ management of the electoral process. For many African countries, voter registration is not an ongoing process. Instead, early in the election cycle EMBs conduct voter registration exercises either to expand the existing voters list or to develop a completely new voter register (as was the case in Nigeria 31 To construct the autonomy index, I normalized each of the three indicators on a 0 to 1 scale. The three indicators were then combined in an additive 0-3 scale which was then normalized on a 0-1 scale. 32 The autonomy index has a Chronbach’s alpha of 0.752 (3 items; (a = 0.752). 33 The capacity index has a Chronbach’s alpha of 0.829 (4 items; (a = 0.829). 259 for the 2007 elections when the registration process began in October 2006). In fact, INEC’s attempts to introduce a new electronic voter registration system in 2006 were beset by numerous technical and human resource challenges which resulted in the disenfranchisement of millions of Nigerians. Citizens and civil society groups chastised the electoral commission for botching the voter registration process. A second way citizens can assess EMB capacity is through the quality of information EMBs provide about elections and voting procedures. Because of the electorates’ low levels of education and inexperience with multiparty elections, EMBs in many emerging countries are mandated to engage in national voter education campaigns. These efforts provide extensive information on various aspects of the electoral process such as the location of the polling stations, how ballots should be cast, and basic regulations and laws guiding the voting process. Having accurate information can make a difference for citizens’ on the day of elections. For instance, many citizens experience difficulties in locating the correct polling station and they will often cast blame on the electoral commission for not providing sufficient information. The third and fourth indicators are more straightforward. Citizen assessments of EMBs often reflect their evaluations of the competence of poll workers. In many respects, the length of time spent voting or the orderliness of the polling station may shed light on extent to which EMB staff are knowledgeable of the basic procedures that guide the election process. Moreover, voters may also associate the capacity of EMBs with the conditions of the polling stations. This is something that is particularly important for the elderly, women with young children, and persons with disabilities who encounter numerous problems accessing polling stations and voting booths. Table 7.11 displays Nigerians’ evaluations of the four indicators of INEC capacity as well as the capacity index. We can infer from these average evaluations that citizens see INEC as 260 most deficient in the capacity of voting facilities and equipment (mean = 0.57) as well as the competence of the polling staff (mean= 0.63). 7.3.6 Variations in Citizens’ Evaluations of INEC Capacity and Autonomy In addition to examining the dimensionality of citizens’ evaluations of INEC performance, I also consider the variations in citizens’ evaluations of INEC autonomy and capacity. Are there any regional variations in citizens’ autonomy/capacity evaluations? Figure 7.2 clearly demonstrates the variation in citizens’ evaluations of INEC autonomy and capacity across Nigeria’s six main geo-political zones. 34 INEC autonomy and capacity are highest for citizens residing in the North Central zone of the country and lowest for those in the South East. The relatively low performance evaluations correspond to the actual conduct of the 2007 elections in the South East including Anambra and Enugu where the most egregious lapses in INEC logistical preparedness and incidents of electoral fraud took place in the various states (EU EOM 2007). What is more interesting is that the gap between autonomy and capacity evaluations is not consistent across the zones. In the North West, which includes states such as Kano and Kaduna, the gap between autonomy and capacity is smallest (0.1), while in the South East the gap is the largest (0.23). Citizens’ evaluations of INEC autonomy and capacity also vary by their partisan affiliations (See Figure 7.3). Those citizens who report being affiliated with the ruling PDP had the highest evaluations of INEC autonomy (0.76) and capacity (0.78). In fact, the gap between autonomy and capacity is inside the margin of error, indicting that for PDP partisans both dimensions are indistinguishable. This is not the case for either supporters of opposition parties 34 By extrapolating the findings to the six geo-political zones the margin of error increases to +/- 4.5%. 261 (such as the All Nigerian Peoples Party or the Action Congress) and non-partisans. Of the two groups of Nigerians, those with no partisan affiliations had the lowest evaluations of autonomy (0.29) and capacity (0.51). Unlike supporters of the ruling party both opposition-partisans and non- partisans were more critical of EMB autonomy. Does education have an impact on how citizens view the two dimensions of EMB performance? Figure 7.4 indicates that differences in educational attainment have a marginally negative effect on citizens’ evaluation of autonomy (difference= 0.06 and statistically significant) but no impact on capacity evaluations (difference = 0). That is to say, educated Nigerians described INEC as having lower levels of autonomy than less educated Nigerians. Turning finally to citizens’ experience with violence. Since 1999, violence has become one of the most pervasive features of Nigerian elections (See Figure 7.5). The results show that citizens across all levels of experience with violence were more critical of INEC autonomy, and citizens with direct experience with violence had the lowest evaluations of EMB autonomy (0.37) and capacity (0.6). Moreover I find that for the victims of violence, the gap between autonomy and capacity is the highest (0.23). With this general understanding of Nigerians attitudes towards INEC, my next step is to examine the statistical relationship between popular EMB performance evaluations and election quality perceptions. 7.4 Consequences of INEC Performance on Nigerians 7.4.1 Are Nigerians’ Evaluations of EMB Performance Associated with Perceptions of Election Quality? 7.4.1.1. Data and Measurement 262 As highlighted in the introduction, a principal goal of this chapter is to examine the relationship between citizens’ evaluations of EMB performance (independent variable) and their perceptions of election quality (dependent variable). I expect citizens’ assessments of EMB performance to be positively associated with their election quality perceptions, especially in the context of emerging democracies such as Nigeria, because EMBs serve an important function during elections. When EMBs perform well, by exercising autonomy and possessing the necessary election-related capacity, they are better able to promote an equitable electoral playing field and to minimize the administrative challenges that plague election management. To test the association between evaluations of EMB performance and perceived election quality, I utilize data from Round 4 of the Afrobarometer survey on Nigeria. 35 The main dependent variable—perceived election quality—captures respondents’ perceptions of the 36 quality of the 2007 elections. Respondents were asked “On the whole, how would you rate the freeness and fairness of the last national election held in 2007?” Potential responses included “Not free and fair”, “Free and fair with major problems”, “Free and fair with minor problems,” and “Completely free and fair”. The original variable is re-scaled into a dichotomous variable with “1” indicating high election quality (citizens reported that the elections were “completely 35 Afrobarometer conducted the survey in Nigeria during May 2008 with a nationally representative sample of 2,408 Nigerians that allows for inferences to the national population based on ±2 margin of error at a 95% confidence level. For more information on the technical details of Afrobarometer see: www.afrobarometer.org. 36 Information on the coding of all variables is included in Appendix A. 263 free and fair” or “free and fair with minor problems”), while “0” denotes low election quality (citizens reported that elections were “free or fair with major problems” or “not free and fair”). 37 Table 7.12 shows the distribution of Nigerians’ perceptions of election quality. As we can see, a plurality of respondents rated the elections as “Not free and fair” (37%), while one-quarter of respondents saw the elections as “Free and fair with major problems” (26%). When these two categories are combined, almost two-thirds of Nigerians (63 percent) offered a negative evaluation of the electoral process. On the other hand, 24 percent of citizens rated the elections as free and fair with minor problems while only seven percent of those interviewed were convinced of the absolute quality of elections. When these two categories are combined, one in three Nigerians (31%) had a positive rating of the elections. Figure 7.6 provides some insight into the variations in citizens’ election quality perceptions across the six geo-political zones. Citizens in the North West displayed the highest election quality perceptions (0.46) while citizens in the South East reported the lowest election quality perceptions (0.17) followed by citizens in the South West (0.26) and South South (0.27). As was indicated in Section 2.3, 37 Although re-scaling election quality as a binary variable may limit the variability in citizens’ evaluations, I use the measure because it provides better interpretation of the results, compared to ordinal logit or probit. All empirical models using perceptions of election quality as the DV have been estimated using both the dichotomous and ordinal measures and there are minimal differences in the substantive results. Results of the ordered logit models are not shown but are available upon request. The decision to dichotomize the measure of election quality from the Afrobarometer survey on the basis of low election quality containing the response categories “Not free and fair” and “Free and fair with major problems” and high election quality containing the response categories “Free and fair with minor problems” and “Completely free and fair” is based on the desire to create equivalence between the Afrobarometer and IFES surveys. In the IFES survey the question on election quality has slightly different response categories (“completely free and fair,” “somewhat free and fair,” “not too free and fair,” or “not at all free and fair”). In my estimation, citizens responding to the question in the IFES survey would more likely associate high quality elections with the first two categories and low quality elections with the second two categories. 264 international and domestic observers recorded the highest instances of fraud, violence, and procedural irregularities in the South East and South South geo-political zones. The main independent variable reflects citizens’ evaluations of EMB performance. I utilize a single item measure of EMB performance that taps into citizens’ evaluation of INEC performance in conducting the April 2007 national elections. 38 Along with the indicator of INEC performance, I include indicators of four other institutions that according to the literature on Nigerian elections play important roles in election administration and could potentially influence Nigerians’ attitudes towards election quality (Adebayo and Omotola 2008). These institutions include the 1) Nigerian Police Force, 2) Nigerian Military, 3) Election Tribunals, and, 4) Economic and Financial Crimes Commission (EFCC). While controlling for citizens’ evaluations of the two main security agencies (Police and Military) as well as the judiciary (Election Tribunals) is intuitive, the inclusion of the EFCC, one of Nigerians’ most prominent anti-corruption agencies, warrants further explanation. The EFCC is a Nigerian law enforcement agency responsible for investigating financial crimes. Established in 2003, the agency became very influential during the 2007 elections as it prosecuted numerous political candidates. Of the four institutions, Nigerians were most dissatisfied with the performance of the police with sixty-two percent expressing dissatisfaction. Nigerians were almost split on the performance of the military (46% dissatisfied versus 44% satisfied), while a minority of Nigerians were unimpressed with the performance of the Election Tribunals (39%) and the EFCC (37%). 38 In the following analysis, I will incorporate data that distinguishes between EMB autonomy and EMB capacity. 265 To capture some of the characteristics of the electoral environment, I control for citizens’ experience with election manipulation. To do this, I incorporate three indicators of perceived electoral manipulation: 1) fear of intimidation, 2) evaluations of freedom to vote, and 3) perceived ballot secrecy. I use an indicator for fear of intimidation and violence that asks respondents the extent to which they personally fear becoming a victim during election campaigns. A second indicator assesses citizens’ perceived ability to vote “without feeling pressured”. Although this question has normally been used as a measure of procedural liberty, I believe that it provides another indicator for the extent of intimidation voters may experience. The third indicator gauges whether citizens believe that their voting decisions are private. Ballot secrecy is a core requirement of Nigerian legal framework and regional and international electoral law. The main expectation is that citizens who fear intimidation and violence, worry about being pressured and doubt the secrecy of the ballot are less likely to have confidence in the quality of elections. 39 A close examination of the survey reveals that over four in ten respondents were significantly fearful of intimidation and violence. While an overwhelming majority of Nigerians (71%) believed that they were free to choose for whom to vote, the results were different for ballot secrecy. Over 40 percent of Nigerians felt that powerful individuals were either “somewhat” or “very likely” to find out how they voted. Furthermore, I account for citizens’ individual attributes. First, I assess citizens’ partisan affiliations through two questions. The first gauges whether citizens are affiliated with a specific 39 One limitation of the indicators of electoral irregularities based on Afrobarometer survey is that they do not make specific reference to citizens’ experience with the 2007 Nigerian elections, but more generally to past election experiences. However, since Afrobarometer conducted the survey in May 2008, approximately one year after the 2007 elections, we can infer that citizens’ responses are based on the 2007 elections along with any previous election experience. 266 political party, and the second asks citizens to identify that party. Based on these two questions, I develop three separate binary indicators of whether citizens report 1) an affiliation with the Peoples Democratic Party (PDP)—the winner of the 2007 presidential election, 2) affiliation with any of the opposition parties, 3) or no affiliation with any political parties (i.e., nonpartisans). My expectation is that citizens affiliated with the PDP are more likely to consider elections free and fair, relative to opposition supporters or non-partisans. According to the Afrobarometer survey, almost half of the respondents (48%) classified themselves as nonpartisans indicating low levels of partisan identification. The remaining respondents were split in their allegiances: 28 percent said they were affiliated with the opposition, while 24 percent reported being close to the ruling PDP. Additionally, I include an indicator of citizens’ level of political participation, whether the respondent voted in the 2007 elections (Voted in 2007 Elections). I proxy citizens’ level of political knowledge through two indicators. The first indicator examines citizens’ frequency of obtaining news through radio, TV and newspaper sources. I also include a measure of respondents’ level of education that ranges from no education (0) to post-graduate education (9). Following recent research in Africa, politically knowledgeable citizens are more likely to have greater insight on what free and fair elections should be and more likely to recognize when elections deviate from these norms (Bratton, Mattes and Gyimah-Boadi 2005). In addition to media exposure and education, I incorporate a composite measure for citizens’ level of political interest, based on two items which measure citizens’ interest in public affairs and citizens’ frequency of discussing politics. I expect politically interested respondents to espouse more favorable election quality opinions because they are more likely to be affiliated 267 with the incumbent political party (Moehler 2009). Lastly, I control for various social and demographic factors including age, gender (Female), and urban/rural residency (Urban Resident). 7.4.1.2. Results Table 7.13 summarizes the results of the logistic regression of citizens’ likelihood to consider elections free and fair given a unit change in main explanatory variables. Additionally, I incorporate predicted-probabilities calculations to highlight the relative substantive effects of the significant variables. 40 Table 7.14 illustrates the change in predicted probability of a respondent perceiving election quality to be free and fair given a minimum to maximum change of a specific independent variable, while holding other variables at their means. 41 As anticipated, the results indicate that citizens’ evaluations of EMB performance are significant and positively associated with perceptions of electoral fairness, after controlling for relevant covariates. In fact, as highlighted in Table 7.14, citizens’ evaluations of EMB performance have the largest substantive association with popular election quality perceptions, when compared to all other significant predictors. When Nigerians are highly satisfied with the performance of INEC, they are 38 percent more likely to consider the 2007 elections free and fair. Similarly, citizens’ satisfaction with Military and EFCC are significant in a statistical and substantive sense. However, evaluations of Police and Election Tribunals were not statistically 40 All predicted probability estimates were calculated on the model in Table 7.13 using Clarify (Tomz et al., 2003). When all variables in the model are held at their mean, the predicted probability that a respondent will have positive evaluations of election quality is 0.3. 41 For the marginal effects calculations, all non-binary independent variables are held at their mean, while Gender is set at Female and Residential Location set to Urban. 268 significant. 42 Regarding perceptions of electoral manipulation, the model shows that those who expressed concerns about being victimized during elections are 10 percent less likely to consider elections free and fair. But the two other indicators of electoral manipulation— perceived ballot 43 secrecy and evaluations of freedom to vote—fail to gain significance in the model. Furthermore, I find that Nigerians who are affiliated with opposition parties and those who identify as non-partisans are nine percent and eleven percent less likely to consider elections free and fair, respectively, when compared to citizens affiliated with the Peoples Democratic Party. Importantly, voting also has a significant impact on election quality judgements. Citizens who reported having voted in the 2007 elections are 4 percent more likely to view elections as free and fair. Neither coefficients for media exposure nor level of education are significant. Although politically interested respondents were more likely to espouse positive election quality attitudes, this association failed to attain statistical significance. Finally, urban residency is associated with increased probability of free and fair perceptions controlling for other factors, while gender and age are insignificant. To summarize, the empirical results indicate that during the 2007 elections, Nigerians were most likely to associate their perceptions of election quality with the performance of INEC, even after controlling for the performance of other election-related institutions, as well as important individual attributes such as partisan affiliation and degree of electoral participation. 42 Given the prominence of the police in elections, the absence of a significant effect on perceptions of election quality is puzzling. In an attempt to understand the finding, I ran subsequent models without the military, and found that citizen evaluations of police performance had a meaningful impact on election quality. 43 To ensure that the effects of Fear of Intimidation did not overshadow the results regarding Secrecy of the Ballot and Freedom to Vote, I estimated another model without Fear of Intimidation, but the coefficients of both variables remained insignificant. 269 A potential limitation of the findings regarding the relationship between EMB performance and election quality perceptions is endogeneity. One may argue that citizens’ EMB performance evaluations are influenced by their perceptions of election quality. I recognize that the causal arrow between EMB performance evaluations and popular election quality perceptions could plausibly run in both directions. However, the specific direction of causation is difficult to disentangle conceptually because during an election period citizens’ judgments of institutional performance influence, and are influenced by, their election quality perceptions. Furthermore, making an empirical assessment of causality is even more problematic because the findings of the previous analyses are based on surveys conducted after the 2007 Nigerian elections. While I acknowledge the limitations that endogeneity poses to the main inferences, empirically resolving the issues associated with endogeneity is beyond the scope of this paper. 7.4.2 Consequences of INEC Autonomy and Capacity on Nigerians’ Election Quality Evaluations The results of the previous statistical analysis reinforce the robust association between EMB performance evaluations and election quality perceptions. However, we can gain a more nuanced understanding of the relationship between EMB performance evaluations and election quality perceptions through a closer examination of various dimensions of EMB performance. With this in mind, this section seeks to answer the following questions. Given an electoral context characterized by widespread irregularities (as was the case in Nigeria during 2007), what dimensions of EMB institutional performance are most important citizens when judging election quality? More specifically, are citizens able to distinguish between EMB autonomy and EMB capacity? Are there conditions under which citizens consider one dimension of EMB performance more important than the other when developing their election quality judgements? 270 To assess the autonomy-capacity distinction, I utilize data from International Foundation of Electoral Systems (IFES) post-election survey conducted in Nigeria less than a month after the 44 2007 elections. I incorporate this survey because it includes a battery of questions that probe citizens’ evaluations of different aspects of the election process, specifically probing dimensions of INEC capacity and autonomy. In examining the relationship between dimensions of EMB performance and election quality perceptions, I follow an empirical strategy similar to that used in the previous analysis. The main variable of interest is perceptions of election quality. IFES asks respondents “In your opinion, how free and fair were the 2007 Presidential and Parliamentary elections? Were the 2007 Presidential and Parliamentary elections completely free and fair, somewhat free and fair, not too free and fair or not at all free and fair?” As done in the previous analysis with Afrobarometer data, I rescale this indicator into a dichotomous variable. The distribution of citizens’ election quality perceptions based on the IFES survey differed slightly from the Afrobarometer survey. A slim majority of Nigerians had a negative rating of the elections when polled in May 2007 (IFES) (“Not free and fair” and “ Free and fair, with major problems”= 51 percent) when compared to 63 percent of Nigerians in May 2008 (Afrobarometer). Moving to the main explanatory variables, I separate INEC performance into two dimensions: 1) autonomy, and 2) capacity, and examine both aspects separately but 44 IFES conducted the survey between May 3 and 10 with a nationally representative sample of 2,416 Nigerians that allows for national inferences based on ±2 margin of error at a 95% confidence level. 271 45 simultaneously. In Section 7.3, I outlined the items and steps in creating the autonomy index and the capacity index. 46 Additionally, I examine the performance of election-related institutions including the police, military and EFCC. 47 IFES also asks slightly different questions regarding electoral malpractice. The item on ballot secrecy is similar to the Afrobarometer survey, but IFES adds three additional questions that gauge experience with intimidation, vote buying, and violence. I expect that experience with intimidation should reduce perceptions of election quality, while the effect of experience with vote buying may be ambiguous given the acceptability of the practice in Nigeria (Bratton 2008). I also include controls for partisanship, political participation, media exposure, political interest and social structure. 7.4.2.1. Results I use a logistic regression model, to examine the effect of EMB autonomy and capacity on Nigerians’ election quality perceptions. The results are shown in Table 7.15. They reveal that Nigerians’ evaluations of INEC autonomy and capacity are both positively correlated with perceptions of election quality. That is, the more Nigerians believe that INEC is autonomous from the control of political elite and has the capacity to effectively administer elections; the 45 The index includes three items that measure respondents’ evaluations of 1) the overall neutrality of INEC, 2) the impartiality of poll workers, and 3) the impartiality of the counting and tabulation processes. The autonomy index ranges from 0 to 1 with increasing levels of perceived autonomy. 46 This index includes four items that measure respondents’ opinions about 1) the voter registration process, 2) the adequacy of information regarding the poll location and voting procedures, 3) the competence of poll workers, and 4) the adequacy of poll facilities. The capacity index ranges from 0 to 1 with increasing levels of perceived capacity. 47 The IFES survey, however, does not include questions examining the performance of Election Tribunals. 272 more citizens regard elections as credible, after controlling for other relevant covariates. We can infer from the findings (mainly the significant coefficients of EMB autonomy and capacity) that both dimensions of INEC performance matter for citizens’ perceptions of election quality. However, do the results of the analysis provide any indication that Nigerians place greater weight on autonomy or capacity when assessing election quality? The predicted probability graphs in Figure 7.7 provide some insight into the differential effects of INEC capacity and autonomy. It highlights the change in predicted probability that citizens will consider elections free and fair, given changes in their evaluations of INEC autonomy or capacity. A close examination reveals that the more positive Nigerians’ evaluations of INEC capacity and autonomy, the more favorable their judgments of election quality. Importantly, however, citizens’ evaluations of INEC autonomy seem to have a greater impact on their judgments of election quality than do their perceptions of administrative capacity. That is, Nigerians who believe that INEC has a high degree of autonomy are 53 percent more likely to consider the 2007 elections favorable, while Nigerians who think that INEC has a high degree of capacity are 32 percent more likely to consider the elections favorably. 48 Based on these findings we can infer that, while both dimensions of institutional performance matter, during the 2007 elections Nigerians paid more attention to INECs institutional autonomy compared to capacity. It might be the case that Nigerians place slightly more weight on INEC’s institutional autonomy because they consider the EMB’s ability to insulate itself from political manipulation the most important function in a country that has been characterized by a history of fraudulent 48 Predicted probability estimates are based on the model in Table 4 and calculated with Clarify (Tomz et al. 2003). For the predicted probability analysis, all non-binary independent variables are held at their mean, while Gender is set at Female and Residential Location set to Urban. 273 elections and a culture of divisive politics. This is not to say that evaluations of capacity are not relevant. On the contrary, when Nigerians’ have low evaluations of both autonomy and capacity the perceived probability of having free and fair elections falls precipitously to 7 percent. Turning now to the control variables, the model indicates that citizens’ affiliated with opposition parties and those who identified themselves as non-partisans were less likely to view the elections in a favorable light, compared to citizens affiliated with the PDP. Furthermore, citizens who experienced intimidation and those who were affected by violence held more negative evaluations of the 2007 elections. In fact, experience with intimidation and violence lowers the probability of free and fair election attitudes by 14 percent and 23 percent respectively; see Table 7.16. But citizens’ experiences with vote buying had no impact on popular election-quality judgments. Similar to the AB findings, voters (relative to non-voters) regarded the elections more favorably. Furthermore, those interested in politics had more positive election quality evaluations, while respondents with greater media exposure were more critical of election quality. In the final analysis, the findings raise the possibility that citizens are able to prioritize the institutional autonomy of EMB performance relative to institutional capacity. While the extent to which I can generalize these findings beyond Nigeria is limited, this research provides a platform for discussing the relative importance of institutional performance dimensions at different stages during the democratization process. For Nigerians during the 2007 elections, the autonomy of the electoral management body is a relatively important pre-requisite for elections to be considered credible. Perhaps Nigeria’s long history of fraudulent and politically manipulated elections has made citizens more attuned to the issue of autonomy, and as a result they are more insistent on election administration that is independent of political interference. 274 This is not to say that EMB capacity is not important, but it may be the case that citizens are more forgiving of lapses in EMB capacity once the autonomy of the body has been established. However, citizens’ tolerance of deficiencies in capacity may not be extended indefinitely. For example, in South Africa citizens have considered the main EMB (Independent Electoral Commission) autonomous for a series of national elections. South Africans are now more likely to focus their criticisms on the problems related to IEC institutional capacity such as the poor quality of voting facilities, long lines during voting, and unavailability of electronic voting procedures (Maphunye, 2010). 7.4.3 Comparing the 2007 and 2011 Elections In addition to examining the relationship between citizens’ evaluations of EMBs and their election quality perceptions during a single election period, the final part of the chapter assesses changes in perceptions of election quality over time. As indicated in the introduction, Nigeria presents researchers with a natural experiment to examine the changes in citizen attitudes following substantial reforms in election administration introduced in the months leading up to the April, 2011 elections. As highlighted in Table 7.7, Nigeria’s election administrative institutions benefitted from several important legal and procedural reforms before the 2011 elections. Some of the most notable changes included greater INEC administrative and financial autonomy, new biometric voter register, and the appointment of a new INEC chairman. If EMBs really matter for citizens, we should expect these changes to be positively associated with citizens’ attitudes towards the electoral commission as well as their perceptions of election quality. I assume that these legislative and procedural reforms represent stimuli (independent variable) that should be associated with various attitudes towards the elections (dependent variable). Furthermore, the 275 aim of the analysis is not to prove a causal relationship but to establish a link between reforms in election administration and citizen election attitudes. I begin by examining the overall changes in citizens’ perceptions of election quality between 2007 and 2011. To do this, I rely on data from the Afrobarometer collected in 2008 and 2012. According to the Afrobarometer, 70 percent of respondents surveyed in 2012 had a favorable rating of the elections (that is they rated the elections as either “free and fair” or “free 49 and fair with minor problems”) when compared to 31 percent of citizens surveyed in 2007. The differences between 2007 and 2011 represent a remarkable 39 percent increase in positive election quality ratings. In the 2012 Afrobarometer survey, respondents were asked to compare the 2011 elections with the 2007. Approximately 2/3 of respondents felt that the elections had improved (“much better” + “better” = 65%), while only 15 percent felt that election quality had declined (“much worse” + “worse”). These findings provide clear evidence that citizens’ attitudes towards election quality improved drastically between 2007 and 2011. Furthermore, we can safely assume that some of these changes in election-quality perceptions are due to perceived improvements in election administration. Instead of simply assuming a link between EMB reform and citizens’ election quality attitudes, it is instructive to examine whether citizens’ views towards INEC have changed between 2007 and 2011. The rationale for asking this question is that changes in citizens’ views toward INEC across 2007 and 2012 would provide additional evidence of the popular salience of electoral administrative reforms. One of the most important reforms of the 2011 election period was the appointment of a new chair of INEC. Following the nomination of Professor Jega on June 10, 2010 Nigerians were asked about their opinion on the nomination in a survey conducted 49 These results are based on the following question: “On the whole, how would you rate the freeness and fairness of the last national election held in 2007?” 276 by NOI Polls. 50 Of those surveyed in August, 80 percent reported having known about the appointment of the new INEC commissioner. 51 Among citizens who were aware of the appointment, 73 percent were confident Jega’s stewardship would enhance INEC’s ability to conduct free and fair elections, while 10 percent felt that the appointment would have no consequences. However, the appointment of Jega on its own, might not be sufficient the sway the opinions of Nigerians who had over the years come to distrust INEC. Toward the end of 2010, INEC began to introduce many of its own reforms that may also have influenced citizens’ evaluations of INEC performance relative to 2007. First, I consider citizens’ overall evaluations of INEC performance across the two election periods. In 2007, 37 percent of citizens rated INEC performance positively (Fairly good: 29%; Very good: 8%) while in 2011 71 percent of Nigerians viewed INEC performance positively (Fairly good: 51%; Very good: 20%). Second, there were dimensions of the electoral administrative reform that may have enhanced citizens’ perceptions of INEC autonomy. For instance, INEC took various steps to create more transparency and reduce electoral fraud on election day. These included the openmodified ballot system of voting and posting of election results at polling stations and collation centers. A closer examination of the data shows that in 2007, 47 percent of Nigerians considered 50 NOI Polls randomly selected respondents for the snap polls from a database of phoneowning Nigerians aged 18 and above. A total of 1024 people took part in the telephone interviews from the 23rd and 25th of August 2010. For a sample of this size, we can say with 95% confidence that the maximum margin of sampling error is ±3 percentage points. NOI polls is affiliated with Gallup international and was founded by Ngozi Okonjo-Iweala — the current Nigerian Minster of Finance. 51 The popularity of the appointment provides some indication that citizens’ were very engaged in the reform process. 277 INEC to be a neutral institution in its operations. 52 Perceptions of INEC neutrality increased after the 2011 elections; with over 2/3 of respondents (68 percent) believing that INEC “performs its duties as a neutral body guided by the law.” 53 Furthermore, Nigerians’ perceptions of the impartiality of the counting and tabulation process were similarly boosted in 2011. Popular satisfaction with one of the most crucial stages of the election process increased from 53 percent of respondents’ in 2007 54 to 73 percent of respondents in 2011. 55 One possible reason for the increase was the decision to post election results at polling stations and collation centers. Another was the decision to use persons with a reputation for impartiality within the constituency to announce the parliamentary results. INEC reforms may have also shaped citizens’ perceptions of EMB capacity. First, INEC developed various new systems to manage the distribution of election materials. Also important, was the utilization of National Youth Service volunteers as temporary polling staff during the elections. According to a senior director of INEC, the main advantage of using NYSC is that they know themselves, they are exuberant, well educated and less liable to be politically manipulated; 52 In 2007 IFES asked respondents to chose the statement closest to their views: 1) “INEC performs its duty as a neutral body guided by in its work by the law”, or 2) “INEC makes decisions which favor particular people or interests.” 53 In 2012 Afrobarometer asked respondents to choose the statement closest to their views: 1) “INEC performs its duties as a neutral body guided only by law”; or 2) “INEC makes decisions that favor particular people, parties or interests.” 54 In 2007 IFES asked respondents how satisfied or dissatisfied they were with the impartiality of the counting, tabulation, and announcement of results during the 2007 presidential and parliamentary elections. 55 In October 2011, the International Republican Institute conducted a nationally representative poll of 3,078 adult Nigerians with a margin of error of +/- 1.8 percent. IRI asked Nigerians, “Overall, how confident are you that the ballots cast in your state were collated without error, interference or tampering by INEC, parties or politicians?” 278 when compared with the adult staff, they are much more reliable. The adult staff [former ad-hoc workers] often give wrong information so it is hard to verify their existence. They cannot be removed from their individual communities and often comply with demands of the local politicians and general public: [they are] more susceptible to manipulation. These changes may have influenced how citizens assessed the dimensions of EMB capacity across the two election periods. Citizens’ satisfaction with the adequacy of voting materials increased by 11 percent between 2007 (61%) and 2011 (72%). Another dimension of EMB capacity is poll worker competence. In 2011, when asked whether polling staff were more competent than 2007 workers, 70 percent of respondents considered the staff in 2011 more competent. This may have reflected Nigerians greater satisfaction and trust in the young college 56 educated graduates who served as NYSC volunteers in 2011. In sum, the comparison of citizens’ opinions on various aspects of the 2007 and 2011 elections indicate that Nigerians enjoyed a freer and fairer electoral experience in 2011. Furthermore, we can infer from the analysis that the improvements in election quality perceptions are associated with the higher levels of popular satisfaction with INEC performance, especially as it has to do with citizens’ evaluation of INEC autonomy and election-day logistical capacity. Finally, the analysis provides some indication that electoral administrative reforms can shape citizens’ views towards electoral commissions and be associated with their election quality perceptions. This was most clearly seen with the appointment of a new and nationally credible electoral commissioner. 56 In 2007, public school teachers were contracted as ad-hoc polling officials. 279 7.5 Conclusion In this chapter, I suggest that assessing variations in citizens’ evaluations of electoral management body performance is important for understanding the relationship between EMBs and citizens’ perceptions of elections, which are the most pivotal event in citizens’ democratic experience. Using recent survey data from elections in Nigeria, I find that citizens’ evaluations of the performance of Independent National Electoral Commission, the main electoral management body in Nigeria, is positively related to perceptions of election quality during the 2007 general elections. Moreover, considering other correlates of citizens’ election quality perceptions, I find that popular evaluations of INEC had the strongest association with election quality perceptions during the 2007 elections. In addition to assessing heterogeneity in citizens’ INEC performance evaluations, the chapter also probes the dimensionality in EMB performance evaluations. Building on one of the main themes throughout the dissertation, I distinguish between EMB autonomy and capacity. Then, I consider whether citizens’ evaluations of dimensions of EMB autonomy and capacity have differential effects on their election quality perceptions. The empirical analysis indicates that, during the 2007 elections, citizens’ evaluations of autonomy and capacity are strongly associated with election quality perceptions. However, citizens seem to place greater weight on their evaluations of INEC autonomy relative to capacity. Finally, because of the important reforms that were implemented in 2010 and 2011 ahead of the April 2011 elections, Nigeria represents an ideal case study of the effect of legal reforms and procedural reforms in election management on citizens’ perceptions of election quality. Comparing survey data on the 2007 and 2011 Nigerian elections, I find that citizens consider the reforms as being instrumental to enhancing their evaluations of INEC performance and overall election quality perceptions. 280 APPENDIX 281 Table 7.1: Nigeria Election Dates (1964-2011) Election Years Presidential 1964 – 1979 11-Aug 1983 6-Aug 1992 – 1993 12-Jun 1999 27-Feb 2003 19-Apr 2007 21-Apr 2011 16-Apr Source: African Elections Database National Assembly 30-Dec 7-Jul 20-Aug 4-Jul – 20-Feb 12-Apr 21-Apr 9-Apr 282 Gubernatorial – – – – – 9-Jan 19-Apr 19-Apr 26-Apr Table 7.2: Nigeria Presidential Election Results (1999-2011) Presidential Candidate Party Number of Votes 1999 Olusegun Obasanjo PDP 18,738,154 Olu Falae AD11,110,287 PDP 2003 Olusegun Obasanjo PDP 24,456,140 Muhammadu Buhari ANPP 12,710,022 Ojukwu Chukwuemeka APGA 1,297,445 Odumegwu 2007 Umaru Musa Yar’Adua PDP 24,638,063 Muhammadu Buhari ANPP 6,605,299 Atiku Abubakar AC 2,637,848 Orji Uzor Kalu PPA 608,803 2011 Goodluck Jonathan PDP 22,495,187 Muhammadu Buhari CPC 12,214,853 Nuhu Ribadu CAN 2,079,151 Ibrahim Shekarau ANPP 917,012 Source: African Elections Database, EU EOM Observer Reports 283 Percentage of Votes 63% 37% 62% 32% 3% 70% 19% 7% 2% 59% 32% 5% 2% Table 7.3: Nigeria National Assembly Election Results (1999-2011) Political Party Senate Seats (109) 1999 People’s Democratic Party (PDP) 59 All People’s Party (APP) 29 Alliance for Democracy (AD) 20 Undeclared 1 2003 People’s Democratic Party (PDP) 76 All Nigeria People’s Party (ANPP) 27 Alliance for Democracy (AD) 6 United Nigeria People’s Party 0 (UNPP) Others 0 2007 People’s Democratic Party (PDP) 87 All Nigeria People’s Party (ANPP) 14 Action Congress (AC) 6 Progressive People’s Alliance 1 (PPA) Others 1 2011 People’s Democratic Party (PDP) 45 Action Congress of Nigeria (ACN) 13 All Nigeria People’s Party (ANPP) 7 Congress for Progressive Change 5 (CPC) Others 4 Source: African Elections Database, EU EOM Observer Reports 284 House of Reps Seats (360) 206 74 68 12 223 96 34 2 4 263 63 30 3 1 123 47 25 30 9 Table 7.4: Nigeria Registration and Turnout (1999-2011) Registered Turnout Invalid Votes Electi # % # % # % 1999 57,938,9 4 30,280,05 5 431,611 1 45 1 2 2 2003 60,823,0 4 42,018,73 6 2,538,24 6 22 3 5 9 6 2007 35,397,5 6 – 5 – – 17 1 7 2011 73,528,0 – 39,469,48 5 1,259,50 3 3 Source: African 40 Elections Database,4EU EOM Observer6Reports 285 Total # 29,848,4 41 39,480,4 89 – 38,209,9 78 % 9 9 4 – 9 7 Table 7.5: Nigeria Election Violence (1999-2011) Election Period Election Fatalities 1999 N/A 2003 105 2007 200 2011 800 Source: African Elections Database, EU EOM Observer Reports 286 Table 7.6: Nigeria Election Petitions and Boycotts (1999-2011) Election Period Election Fatalities 1999 N/A 2003 560 2007 1250 2011 400 Source: African Elections Database, EU EOM Observer Reports 287 Table 7.7: Nigeria EMB Reform (1999-2011) Election Period Electoral Reform 1999 Constitution; Decree N. 17 of 5 August 1988 Establishment of INEC: Powers, functions, composition of INEC 2003 Electoral Act 2003 (2001,2002) Observers and party agents to observe most aspects of the electoral process; the counting takes place at the polling stations Computerized Voters Roll Computerized Voter Register using Optical Mark Recognition Distribution of VR Cards 2007 Electoral Act 2006 INEC appoints the INEC Secretariat Creation of fund for INEC financial independence New procedures for voter registration Responsibility to promote voter and civic education Power to prosecute electoral offenders Biometric Voter Registration Electronic Voter Registration based on Direct Data Capture machines 2011 2010 Amendment to the 1999 Constitution INEC sets the date and order of elections; financial autonomy of INEC Bars political representatives from being appointed as INEC commissioners INEC independence from political control; INEC authority to promote internal- party democracy Electoral Act 2011 (2010) Detailed regulations for political party primaries Posting election results at polling stations EMB Procedures Modified Open Ballot System National Youth Service Corps as temporary polling staff 288 Table 7.8: INEC Chairpersons (1999-2011) INEC Chairpersons Hon. Justice Ephraim Akpata Dr. Abel Guobadia Prof. Maurice Iwu Prof. Attahiru Jega Source: Jinadu (2011) Tenure 1998-2000 2000-2005 2005-2010 2010-present 289 Table 7.9: Nigerians’ Evaluation of INEC Performance (2007-2011) Evaluations of INEC 2007 2011 Performance Elections Elections Very poor 33 9 Fairly poor 25 18 Fairly good 29 51 Very good 8 20 Don’t know 6 2 Total 2408 2,400 Question: How would you evaluate the Independent National Electoral Commission’s performance in conducting the April 2007/2011 national elections? Source: Afrobarometer Round 4 and 5 290 Table 7.10: INEC Autonomy (2007) Dimensions of Autonomy INEC neutrality Impartiality of counting and tabulation of poll workers Impartiality INEC autonomy index Mean Autonomy Evaluations 0.43 0.51 0.58 0.50 291 Observation s 2242 2252 2253 2055 Table 7.11: INEC Capacity (2007-2011) Dimensions of Capacity Adequacy of election Competence of poll staff Voter registration Information on voting INEC capacity index Mean INEC Capacity Evaluations 0.58 0.63 0.69 0.70 0.65 292 Observatio ns 2272 2265 2343 2334 2235 Table 7.12: Nigerians Election Quality Perceptions (2007) Original Coding % Dichotomous Coding % Not free & fair 36 Not F&F + F&F, major problems 6 F&F, major problems 26 F & F, minor problems 24 F&F, minor problems + 31 Completely F & F 7 Completely F&F Do not understand questions 1 Missing variables 6 Don’t know 5 Total 10 10 Question: On the whole, how would you rate the freeness and fairness of the last national elections held in April 2007/2011? Source: Afrobarometer Round 4 293 Table 7.13: Sources of Citizens’ Perceptions of Election Quality (2008: Afrobarometer) Variables Coefficient (Std. Err.) Institutional Performance Evaluations INEC 0.174** (0.085) Police 0.138 (0.102) Military 0.280 (0.094) EFCC 0.401** (0.102) Election Tribunal 0.053 (0.065) Partisanship (reference PDP partisans) Opposition Partisans -0.538** (0.170) Non-partisans -0.581** (0.154) Electoral Irregularities Fear of Intimidation -0.233** (0.063) Freedom to Vote 0.055 (0.079) Ballot Secrecy 0.008 (0.073) Participation and Engagement Voted in 2007 Elections 0.256 (0.167) Political Sophistication Index of Political Interests 0.228 (0.281) Index of Media Exposure -0.508 (0.383) Social Structure Level of Education -0.041 (0.040) Age -0.003 (0.006) Female 0.183 (0.140) Urban Resident 0.458** (0.167) Intercept -2.537** (0.483) N 1511 Log-likelihood -743.691 Chi-Squared (22) 343.74 Pseudo R-squared 0.236 Note: DV Perceptions of Election Quality: Nigeria Afrobarometer Round 4 (2008). Standard errors are clustered at the enumeration area level (primary sampling units). Significance levels: † : 10% ,* 5; **: 1%. 294 Table 7.14: Predicted Probability of Free and Fair Election Quality Perceptions (2008: Afrobarometer) Significant Predictors Change in Predicted Probability 5% C.I 95% C.I Urban Residents -0.059 -0.12 -0.019 Partisanship -0.099 -0.15 -0.034 (Opposition)(NonPartisanship -0.108 -0.171 -0.048 Partisans) Fear Intimidation -0.106 -0.178 -0.043 Performance of 0.134 0.048 0.235 Military Performance of EFCC 0.187 0.11 0.278 Performance of INEC 0.38 0.27 0.491 Note: Predicted probability estimates are based on Table 7.13 and calculated with Clarify (Tomz et al., 2003). For the predicted probability analysis, all non-binary independent variables are held at their mean, while Gender is set at Female and Residential Location set to Urban. (n=1511). 295 Table 7.15: Sources of Citizens’ Perceptions of Election Quality (2007: IFES) Variables Coefficient (Std. Err.) Institutional Performance Evaluations EMB (INEC) Autonomy Index 3.16** (0.313) EMB (INEC) Capacity Index 1.728** (0.472) Police 0.213* (0.099) Military -0.098 (0.106) EFCC 0.141 (0.099) Partisanship (reference PDP partisans) Opposition Partisans -1.391** (0.215) Non-partisans -1.305** (0.368) Electoral Irregularities Experience w/ Intimidation -0.776** (0.241) Experience w/ Vote Buying 0.154 (0.215) Experience w/ Violence -0.605** (0.125) Participation and Engagement Voted in 2007 Elections 0.542* (0.226) Political Sophistication Index of Political Interests 0.817† (0.481) Index of Media Exposure -0.927* (0.523) Social Structure Level of Education 0.041 (0.037) Age 0.001 (0.006) Female -0.164 (0.158) Urban Resident 0.263 (0.207) Intercept -2.450** (0.659) N 1479 Log-likelihood -541.785 Chi-Squared (22) 403.08 Pseudo R-squared 0.473 Note: DV Perceptions of Election Quality: Nigeria Afrobarometer Round 4 (2008). Standard errors are clustered at the enumeration area level (primary sampling units). Significance levels: † : 10% ,* 5; **: 1%. 296 Table 7.16: Predicted Probability of Free and Fair Election Quality Perceptions (2007: IFES) Change in 5% C.I 95% C.I Predicated Probability Voted in 2007 Elections 0.121 0.028 0.229 Political Interests 0.127 0.013 0.266 Media Exposure -0.171 -0.333 -.009 Experience w/ Intimidation -0.173 -0.068 -0.275 Experience w/ Violence -0.267 -0.168 -0.361 Partisanship (Non-0.313 -0.164 -0.445 Partisans) Partisanship (Opposition) -0.328 -0.248 -0.405 INEC Capacity Index 0.369 0.175 .548 INEC Autonomy Index 0.591 0.478 0.697 Note: Predicted probability estimates are based on Table 7.15 and calculated with Clarify (Tomz et al., 2003). For the predicted probability analysis, all non-binary independent variables are held at their mean, while Gender is set at Female and Residential Location set to Urban. (n=1511). 297 Table 7.17: Nigerians Election Quality Perceptions (2007 vs. 2011) by Geo-Political Zones Geopolitical Zones 2007 Elections 2011 Elections North Central 0.42 0.53 North East 0.42 0.48 North West 0.46 0.74 South East 0.17 0.79 South South 0.27 0.76 South West 0.26 0.90 Question: On the whole, how would you rate the freeness and fairness of the last national elections held in April 2007/2011? Source: Afrobarometer Round 4 and 5 298 Figure 7.1: Percentage of Citizens who Trust INEC “A lot or Somewhat” (1999-2011) 33% 2012 (2011 Elections) 28% 2008 (2007 Elections) 21% 2005 (2003 Elections) 62% 1999 (1999 Elections) 0% 10% 20% 30% 40% 50% 60% 70% Question: “How much do you trust the Independent National Electoral Commission (INEC), or haven’t you heard enough about them to say?” Source: Afrobarometer Nigeria Round 1,3,4 and 5. 299 Figure 7.2: INEC Autonomy/Capacity by Geo-Political Zones (2007) 0.39 South West 0.66 0.50 South South 0.25 South East 0.65 0.48 0.57 North West 0.59 North East 0.71 0.63 North Central 0.00 0.67 0.10 0.20 0.30 INEC Autonomy Index 0.40 0.50 0.60 INEC Capacity Index 300 0.70 0.75 0.80 Figure 7.3: INEC Autonomy/Capacity by Partisanship (2007) 0.76 PDP partisans 0.78 0.38 Opposition-partisans 0.61 0.29 Non-partisans 0.51 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 INEC Autonomy Index INEC Capacity Index 301 Figure 7.4: INEC Autonomy/Capacity by Education (2007) 0.45 High Education 0.66 0.51 Low Education 0.66 0.00 0.10 0.20 INEC Autonomy Index 0.30 0.40 0.50 INEC Capacity Index 302 0.60 0.70 Figure 7.5: INEC Autonomy/Capacity by violence Experience (2007) 0.37 Personally expereinced 0.60 0.43 Personally heard 0.62 0.57 No violence 0.00 0.70 0.10 0.20 INEC Autonomy Index 0.30 0.40 0.50 0.60 INEC Capacity Index 303 0.70 0.80 Figure 7.6: Nigerians Election Quality Perceptions by States (2007) North West 0.46 North Central 0.42 North East 0.42 Mean 0.33 South South 0.27 South West 0.26 South East 0.00 0.17 0.10 0.20 0.30 304 0.40 0.50 Figure 7.7: Impact of INEC Autonomy and Capacity on Predicted Probability of Free and Fair Election Quality Perception 305 CHAPTER 8 CONCLUSION In this dissertation, I have examined the critical, but understudied role of electoral management bodies in emerging democracies. The main finding is that EMBs play a crucial role in the democratization process by shaping the behavior and attitudes of political elites and everyday citizens during various stages of the election. I suggest that EMBs may represent instruments of democracy that enhance the procedural legitimacy and substantive uncertainty of elections by effectively managing various dimensions of the election process and holding political elites accountable to the rules of the electoral game. The research highlights the different ways in which citizens, political elites, and members of international community compete over the rules that shape the design of EMBs (Chapter 3). In countries such as Nigeria, citizens and opposition parties have staged protests to demand more independent and effective EMBs. In other countries, such as Zimbabwe, international and regional actors such as the African Union (AU) and Southern African Development Community (SADC) have intervened to ensure that regulations such as the posting of election results at polling stations are included within the legal framework and effectively enforced during elections. The international community has also pumped millions of dollars into enhancing the capacity of EMBs to conduct technologically sophisticated processes such as biometric voter registration operations similar to those developed in Kenya during for 2013 elections. The dissertation also highlights the important consequences of EMB performance on political attitudes and behavior. For instance, I show how Ghanaian politicians’ positive evaluations of election quality (2000-2008) are directly related to the high levels of trust they have in the Electoral Commission. Moreover, in South Africa the Independent Electoral 306 Commission’s dedication to making the voting process as error free as possible has endeared the commission to a majority of South Africans and earned the commission an international reputation for effective election management. By contrast, the legacy of executive manipulation of the Zimbabwean Electoral Commission has made opposition parties and civil society groups very wary of the commissions’ ability to organize credible elections despite important reforms in 2009. By highlighting the role of EMBs, my research seeks to make significant additions to the newly emerging literature on electoral governance and also to expand the research on political institutions in emerging democracies. In particular, I show how in some countries EMBs have been the thread that has preserved democratic stability during tumultuous transitions and functioned as a catalyst for regime legitimacy and democratic deepening. In other countries, the endemic shortage of resources and consistent political manipulation has undermined the performance of EMBs and prevented the emergence of EMBs that can promote democratic elections. Finally, I show that EMBs, such as the INEC in Nigeria, can be effectively reformed even after years of mismanagement and perceived political partiality. This possibility bodes well for other countries in Africa, such as Zimbabwe and Kenya, that have embarked on comprehensive reform of election administration. 8.1 Summary of Findings My research formally defines electoral management bodies (EMBs) as institutions with the responsibility to organize, supervise, and adjudicate some or all of the stages of the election process (Mozaffar 2002; Lopez-Pintor 2000; Wall et al. 2006). I make a distinction between the formal rules that characterize the responsibilities of the EMBs and the actual performance of the institution during the various stages of the election process. EMB performance, I suggested, is 307 structured by the effective implementation and enforcement of formal rules, the pervasiveness of informal rules, and the dynamics of the wider politico-institutional context. Moreover, I conceptually disentangled two dimensions of EMB performance and design—EMB autonomy and capacity—and suggested that these dimensions may be determined by different factors and have distinct impacts on political attitudes and behavior. I probed the significance of EMBs for democratic development from two main perspectives. First, in Chapter 3, I examined the determinants of the design of electoral management bodies. Second, in Chapters 4 through to 7, I explored the consequences of EMB design and performance on elites and citizens’ electoral attitudes and behavior using a variety of methods. More specifically, in Chapter 3 I argued that election administrative design is determined by strategic and dynamic interaction among domestic elites (incumbent and opposition) and international actors. I empirically assessed the framework, with a comparative study of election administrative reform in Ghana, Nigeria, South Africa, and Zimbabwe from 1990-2008. The main findings indicate that EMBs are likely to have greater levels of formal autonomy and capacity when opposition parties and the international donor community exert credible and enduring pressure on the incumbent regime. In Ghana and South Africa, the dominance of the opposition and the international community helped to forge the institutional autonomy and capacity of EMBs in both countries. In contrast, President Mugabe’s strong legal and coercive powers in Zimbabwe made the prospect of reform difficult for the burgeoning opposition. More recently, however, both domestic and international pressure have slowly chipped away at Mugabe’s dominance and enhanced the prospects of greater autonomy and capacity for the Zimbabwe Electoral Commission (ZEC). 308 Chapter 4 probed the consequences of EMB design for the election quality evaluations of elites and citizens. Using original data on different attributes of EMB design, including the de jure autonomy and de jure capacity, I found that EMB design does not explain the cross- time and cross-national variation in the freeness and fairness of elections from the vantage point of both elites and everyday citizens in 20 African countries. I also explained why this finding should not be surprising: in many countries in Africa, EMB rules are weakly enforced, informal rules such as corruption take precedence, and the political and institutional context are unsupportive. Because EMB design does not help to explain differences in election quality, in Chapters 5, 6, and 7, I examined the effect of EMB performance (de facto autonomy and capacity). To measure EMB performance in Chapters 5 and 6, I used an indicator from the Quality of Elections Dataset that reflects the autonomy and capacity of EMBs during different stages of the election process (Kelley 2012). The findings provided strong and consistent support for the positive effect of EMBs on various dimensions of election quality. To be clear, the results showed that when EMBs perform well, political elites are more likely to consider elections free and fair, accept the results of the election and turnover in political leadership is more likely. On the other hand, citizens are more likely to report that elections provided avenues of public participation and electoral competition when EMBs display political autonomy, professionalism, and administrative efficiency. Chapter 7 further emphasized the significant effect of EMB performance on citizens’ election quality attitudes in Nigeria between 1999 and 2011. I relied on data capturing citizens’ own evaluations of EMB performance from public opinion surveys conducted by the Afrobarometer and the International Foundation of Electoral Systems. These surveys gauged the 309 de facto autonomy (such as perceived impartiality of poll workers) and de facto capacity (such as the perceived effectiveness of the voter registration process) of EMBs. Not surprisingly, Nigerians were more likely to associate their perceptions of election quality with the performance of the Independent National Election Commission (INEC). Furthermore, during the 2007 elections Nigerian citizens seemed to place greater weight on their evaluations of INEC autonomy relative to capacity. Finally, the numerous reforms implemented in 2010 tremendously improved Nigerians’ evaluations of INEC performance when compared to 2007. 8.2 Theoretical Implications These findings have important theoretical implications for the scholarship on democratization and political institutions. First and foremost, the findings highlight the need for greater focus on the role of EMBs in the literature on democratization. Traditionally, the democratization literature has vigorously debated the importance of state institutions such as the legislature, judiciary, and the bureaucracy in increasing the prospects of democratic consolidation and making sure that democracy is seen as the only game in town (Geddes 1999; Acemoglu and Robinson 2006; Przeworski 2000). However, my research has demonstrated that EMBs are also crucial to understanding the quality of elections. Since free and fair elections represent a necessary (but not sufficient) condition for democratic consolidation (Dahl 1971; Bratton 1998), more focused study of EMBs will augment the complexity of theories that seek to assess the determinants of democratic consolidation. While this research is not the first to identify EMBs as “islands of excellence” (Pastor 1999; Schedler et al. 1999) that affect the quality of elections and the democratic process more generally, I have been able to show how these dynamics are manifested in multiple regime contexts over different phases of the democratization process. This has given me more leverage 310 to systematically address some important questions, such as the extent to which EMB performance structures the attitudes and behavior elites and citizens (Chapters 5, 6, and 7). Existing studies have considered answering this question critical to understanding the dynamics of election quality evaluations (Birch 2008; Rosas 2010). However, the existing literature has not provided an adequate answer using cross-sectional and cross-time analysis. So my research represents one of the first efforts to fill this gap in the literature. Nonetheless, there are other important lines of inquiry that still remain unanswered, such as the impact of EMB performance on citizens’ electoral participation. EMBs may also help strengthen regime legitimacy in transitioning democracies or hybrid regimes where elites and citizens (especially those affiliated with the opposition) place less faith in the legitimacy of institutions such as legislatures, judiciaries, and bureaucracies that are tainted by incumbent abuse of power. My efforts to conceptually distinguish between EMB autonomy and capacity and empirically test the dimensionality have clear implications for the study of political institutions in emerging democracies. By probing the multi-dimensionality of EMB institutional performance (de facto autonomy and capacity), I have developed a more nuanced analysis of the causes and consequences of EMB performance that has been absent from the existing literature because of the exclusive focus on EMB autonomy. Specifically, I show that in some countries EMBs have disproportionate level of autonomy and capacity. This disproportionality may reflect political actors’ strategic decisions to pursue one dimension at the expense of another. For instance, I showed that in Zimbabwe (2000 and 2002) the incumbents’ ability to expand the capacity of election administration while blocking changes in autonomy provided the incumbent 311 with the ability to more effectively control election administration and devise complex methods of rigging the results. Furthermore, I demonstrate how citizens and elites respond differently to either autonomy or capacity. A perfect example of this is highlighted in Chapter 7. The insights gained from disentangling dimensions of EMB performance can be applied to the study of other democratic institutions such as the judiciary, legislature, and anti-corruption agencies. Probing the relevant dimensions of institutional performance for institutions like these will help in building more robust theories about their effect on political behavior. 8.3 Policy Implications Beyond theoretical implications, the findings of the research have important consequences for the policy-making community. Particularly important to policy-makers are the findings that EMB performance matters more than EMB design in shaping the political attitudes and behavior. This finding cautions against an exclusive focus on changes in EMB design to improve the performance of the institution. In particular, it indicates that EMB formal reforms are not “silver bullets” for ensuring improved EMB performance and the credibility of elections. Instead domestic policymakers and the international donor community should take a more comprehensive approach that highlights all the factors that influence EMB performance such as the enforcement of formal rules, the effect of informal rules, and the larger institutional context. For instance, opposition and civil society actors in Zimbabwe demand greater mechanisms to ensure that the incumbent will play by the existing rules of the game. Instead of merely changing EMB design, Zimbabwean policymakers and the international community need to focus more on efforts to expand the capacity and professionalism of the judiciary, security agencies, and the 312 media. Hopefully, these changes could provide a more supportive context for ZEC to operate more effectively and to check executive abuses of power. Another major finding, is that perceptions of EMB performance matter as much as their actual performance. Policy-makers, especially EMB officials, need to focus on improving the public image of EMBs through greater transparency and public education campaigns. Enhancing voters’ knowledge about the election-related roles of the EMB and the complexity of election administration can help build the popular legitimacy of the institution. In Ghana and South Africa reputation building had been a feature of their electoral commissions. This was especially the case during its early stages of the institutions’ development and continues to be a primary focus even in recent elections. In contrast, commentators see the inability of ZEC to do this in preparation for the upcoming 2013 elections to be a foreboding sign of its inability to enhance the process. 8.4 Research Limitations and Opportunities for the Future Despite these theoretical and policy contributions, there are important limitations in the dissertation that warrant further attention. The first shortcoming is related to findings from the small-n qualitative study in Chapter 3 and the extent to which the main inferences can be generalized to Africa and other emerging democratic contexts. One reason for the limited generalizability is that I limited the cases to former British colonies, to ensure parsimonious theory, and did not consider the effects of different colonial legacies on the type of EMB design. For instance, former French colonies such as Benin and Togo tend to have less nominally independent EMBs because of their rich legacy of professional bureaucracies. A second reason is that the qualitative research design limits the extent to which I can effectively control for other confounding factors and alternative explanations. As a result, I take a very cautious approach to 313 making inferences to other emerging democracies and limit the scope conditions to African countries that share similarities to Ghana, Nigeria, South Africa, and Zimbabwe. My hesitation to applying the findings of the study more generally to other emerging democracies provides an opportunity for future research. I suggest developing a mixed-method research design that integrates a large-n statistical analysis based on data that includes some of the characteristics that have been identified as being central to the variations in design in the qualitative study (in Chapter 3). A second potential limitation relates to the findings concerning the impact of EMB performance on political attitudes and behavior (Chapters 5-7). It is possible that the validity of the results is called into question because of endogeneity between political actors’ evaluations of EMB performance and their electoral attitudes and behaviors. Generally, concerns about endogeneity arise because EMBs play such an important role in various stages of the election process. As a result, it is difficult to empirically decipher whether political actors’ evaluation of EMB performance is based on their perceptions of election quality or vice-a-versa. This is particularly problematic for the analysis in Chapter 7. The research design does not provide adequate leverage to establish whether citizens wait until after elections and use their evaluations of election quality to rate the performance of EMBs. While I have incorporated various methods to identify, and in some cases address, the attendant bias attributable to endogeneity, these efforts are far from perfect. 1 In future research, I intended to more effectively address the problem of endogeneity. One potential solution involves the development of a panel survey that tracks the views of a cohort of citizens at various moments during the pre-election, election and post-election stages. 1 For instance, in Chapter 7, I use a research design that measures changes in election quality before and after the institutions of EMB reforms during the 2010 election period. 314 The survey will gauge variations in popular evaluations of election administration, election quality, and political preferences. Importantly, the survey will examine whether there are significant changes in perceptions of election quality for citizens whose expectations of their party’s victory (loss) in the pre-election period were not realized after the announcement of election results. The most ideal context for the project is a country with competitive national and parliamentary elections. 315 BIBLIOGRAPHY 316 BIBLIOGRAPHY Adebayo, P. and S. Omotola. 2007. “Public perception of the 2007 Nigerian general elections.” Journal of African Elections 6(2): 201–219. Aiyede, E. 2012. “The politics of electoral reform in Nigeria, 2007-2011: Editorial.” Journal of African Elections 11(1): 1-7. Akhaine, S. O. 2012. “Nigeria’s 2011 elections: The ‘crippled giant’ learns to walk?” African Affairs 110(441):649–655. Alabi, M. O. A. 2009. “Electoral Reforms and Democratic Consolidation in Nigeria: The Electoral Act 2006.” CEU Political Science Journal (02):278. Alvarez, R. M. and T. E. Hall. 2006. “Controlling democracy: The principal-agent problems in election administration.” Policy Studies Journal 34(4):491–510. Alvarez, R. M., T. E. Hall and M. H. Llewellyn. 2008. “Are Americans confident their ballots are counted?” The Journal of Politics 70(03):754–766. Anderson, C. J. and A. J. LoTempio. 2002. “Winning, losing and political trust in America.” British Journal of Political Science 32:335–351. Anglin, D. G. 1995. “International monitoring of the transition to democracy in South Africa, 1992-1994.” African Affairs 94(377):519. Ayee, J. R. A. 1997. 16(3):416–427. “The December 1996 general elections in Ghana.” Electoral Studies Ayee, J. R. A. 1998. “Election Management and Democratic Consolidation: The Election Commission in Ghana.”. Banducci, S. A. and J. A. Karp. 2003. “How elections change the way citizens view the political system: campaigns, media effects and electoral outcomes in comparative 317 perspective.” British Journal of Political Science 33(03):443–467. Barkan, J. D. 2009. Legislative power in emerging African democracies. Boulder: Lynne Rienner Publishers. Bayart, J. 1993. The state in Africa : the politics of the belly. London ; New York, N.Y.: Longman. Beaulieu, E. and S. D. Hyde. 2009. “In the Shadow of Democracy Promotion.” Comparative Political Studies 42(3):392. Birch, S. 2008. “Electoral Institutions and Popular Confidence in Electoral Processes: A CrossNational Analysis.” Electoral Studies 27(2):305–321. Birch, S. 2011. Electoral malpractice. Oxford University Press. Boahen, A. 1995b. “A note on the Ghanaian elections.” African Affairs 94(375): 277–280. Boda, M. D. 2005. “Reconsidering the free and fair question.” Representation 41(3): 155–160. Boone, C. 2003. Political topographies of the African state: territorial authority and institutional choice. Cambridge studies in comparative politics. Cambridge, U.K. ; New York: Cambridge University Press. Booysen, S. 2009. “The presidential and parliamentary elections in Zimbabwe, March and June 2008.” Electoral Studies 28(1):150–154. Bowler, S. and T. Donovan. 2012. “The Limited Effects of Election Reforms on Efficacy and Engagement.” Australian Journal of Political Science 47(1):55–70. Bratton, M. 1994. Engaging the State: Associational Life in Sub-Saharan Africa. In State power and social forces : domination and transformation in the Third World, ed. Joel S. Migdal, Atul Kohli and Vivienne Shue. Cambridge [England] ; New York: Cambridge University Press. Bratton, M.. 1998. “Second Elections In Africa.” Journal Of Democracy . 318 Bratton, M. 2007. “Formal Versus Informal Institutions In Africa.” Journal of Democracy 18(3):96–110. Bratton, Michael. 2013. Voting and Democratic Citizenship in Africa. Lynne Rienner Publishers. Bratton, M. 2008. “Vote buying and violence in Nigerian election campaigns.” Electoral Studies 27(4):621–632. Bratton, M. and E. C. C. Chang. 2006. “State building and democratization in sub-Saharan Africa: forwards, backwards, or together?” Comparative Political Studies 39(9):1059. Bratton, M. and E. Masunungure. 2008. “Zimbabwe’s long agony.” Journal of Democracy 19(4):41–55. Bratton, M. and N. van de Walle. 1997. Democratic experiments in Africa: regime transitions in comparative perspective. Cambridge studies in comparative politics. Cambridge, U.K. ; New York: Cambridge University Press. Bratton, M., R. Mattes and E. Gyimah-Boadi. 2005. Public opinion, democracy, and market reform in Africa. Cambridge studies in comparative politics. Cambridge, UK ; New York: Cambridge University Press. Brownlee, J. 2007. Authoritarianism in an age of democratization. New York: Cambridge University Press. Bunce, V. J. and S. L. Wolchik. 2009. “Defeating Dictators: Electoral Change and Stability in Competitive Authoritarian Regimes.” World Politics 62(01):43–86. Capoccia, G. and D. Kelemen. 2007. “The study of critical junctures: Theory, narrative, and counterfactuals in historical institutionalism.” World Politics 59(03):341–369. Carey, J. M. 2000. “Parchment, equilibria, and institutions.” Comparative Political Studies 33(6-7):735–761. 319 Carothers, T. 1999. Aiding democracy abroad: the learning curve. Carnegie Endowment. Carpenter, D. 2001. The forging of bureaucratic autonomy : reputations, networks, and policy innovation in executive agencies, 1862-1928. Princeton studies in American politics. Princeton, N.J.: Princeton University Press. Cho, W. and M. Bratton. 2006. “Electoral institutions, partisan status, and political support in Lesotho.” Electoral Studies 25(4):731–750. Cho, W. 2010. Do Electoral Systems and Ethnic Diversity Matter?” Comparative Political Studies 43(12):1650–1674. Chu, Y., M. Bratton, M. Lagos, S. Shastri and M. Tessler. 2008. “Public opinion and democratic legitimacy.” Journal of Democracy 19(2):74. Clinton, H. 2011. “Election in Nigeria.” CODEO. 2012. Elections.” “Preliminary Statement of Ghana’s 7 to 8, 2012 Presidential and General Collier, P. and P. C. Vicente. 2010. “Violence, bribery, and fraud: the political economy of elections in Sub-Saharan Africa.” Public Choice pp. 1–31. Committee, Electoral Reform. 2008. “Report of the Electoral Reform Committee of Nigeria.” Conroy-Krutz, J. 2009. Political information and electoral behavior in Sub-Saharan Africa PhD thesis Columbia University. Constitution of Nigeria. 1999. Constitution of Zimbabwe. 2000. Constitution of Zimbabwe. 2009. Constitution of Zimbabwe Amendment (No.17) Act. 2005. Constitution of Zimbabwe Amendment (No.18) Act. 2007. 320 Cox, G. W. 1997. Making votes count : strategic coordination in the world’s electoral systems. Political economy of institutions and decisions. Cambridge ; New York: Cambridge University Press. Cukierman, A., S. B. Webb and B. Neyapti. 1992. “Measuring the Independence of Central Banks and its Effects on Policy Outcomes.” World Bank Economic Review 6(3):353–398. Dahl, R.. 1971. Polyarchy; participation and opposition. New Haven,: Yale University Press. Debrah, E. 2011. “Measuring Governance Institutions’ Success in Ghana: The Case of the Electoral Commission, 1993-2008.” African Studies 70(1):25–45. Diamond, L. 1994. “Toward democratic consolidation.” Journal of Democracy 5(3):4–17. Diamond, L. 2002. “Thinking about hybrid regimes.” Journal of Democracy 13(2):21–35. Downs, A. 1957. An economic theory of democracy. New York,: Harper. Economist. 2011. “Gambling on the world’s most expensive voting system has paid off.” EISA. 2008. “EISA election observer mission report: The Zimbabwe harmonised elections of 29 March 2008, presidential, parliamentary and local government elections.”. Eisenstadt, T. 2004. “Catching the state off guard - Electoral courts, campaign finance, and Mexico’s separation of state and ruling party.” Party Politics 10(6):723–745. Eisenstadt, T. 2002. “Measuring electoral court failure in democratizing Mexico.” International political science review 23(1):47–68. Elklit, J. and A. Reynolds. 2005a. “Judging elections and election management quality by process.” Representation 41(3):189–207. Elklit, J. and A. Reynolds. 2000. “The Impact of Election Administration on the Legimacy of Emerging Democracies: A New Research Agenda.” Working Paper 281. Elklit, J and A. Reynolds. 2005b. “A Framework for the Systematic Study of Election Quality.” Democratization 12(2):15. 321 Engerman, S. L. and K. L. Sokoloff. 2008. “Debating the role of institutions in political and economic development: Theory, history, and findings.” Annual Review of Political Science 11:119–135. Englebert, Pierre. 2000. State legitimacy and development in Africa. Boulder, Colo.: Lynne Rienner Publishers. EU EOM. 2003. “Nigeria Final Report: Presidential, National Assembly Elections and Gubernatorial Elections 12 April - 19 May 2003.” European Union Electoral Observation Mission to Nigeria 2003. EU EOM. 2007. “Nigeria Final Report: Presidential and National Assembly Elections, 21 April 2007.” European Union Electoral Observation Mission to Nigeria 2007. EU EOM. 2011. “Nigeria Final Report: Presidential and National Assembly Elections: European Union Election Observation Mission.” European Union Electoral Observation Mission to Nigeria 2011. Evrensel, A. 2010. Voter Registration in Africa: A comparative analysis. EISA. Fick, G. 1998. The South African Electoral Commission: the Old and the New. Johannesburg: Electoral Institute of South Africa. Fukuyama, F. 2013. “What Is Governance?” Governance. Gandhi, J. 2008. “Coordination among Opposition Parties in Authoritarian Elections.”. Gazibo, M. 2006. “The Forging of Institutional Autonomy: A Comparative Study of Electoral Management Commissions in Africa.” Canadian Journal of Political Science 39(3):22. Geddes, B. 1994. Politician’s dilemma: building state capacity in Latin America. California series on social choice and political economy ; Berkeley: University of California Press. Goodwin-Gill, G. 1994. Free and fair elections : international law and practice. Geneva: InterParliamentary Union. 322 Gyimah-Boadi, E. 1999. Institutionalizing Credible Elections in Ghana. In The self-restraining state : power and accountability in new democracies, ed. Andreas Schedler, Larry Diamond and Marc Plattner. Lynne Rienner Publishers. Gyimah-Boadi, E. 2009. “Another step forward for Ghana.” Journal of Democracy 20(2):138– 152. Hale, H. 2011. “Formal constitutions in informal politics: Institutions and democratization in post-Soviet Eurasia.” World Politics 63(04):581–617. Hale, K. and C. D. Slaton. 2008. “Building Capacity in Election Administration: Local Responses to Complexity and Interdependence.” Public Administration Review 68(5):839–849. Hall, P. A. and R. C. R. Taylor. 1996. “Political science and the three new institutionalisms.” Political Studies 44(5):936–957. Hall, T. E., J. Quin Monson and K. D. Patterson. 2009. “The human dimension of elections: How poll workers shape public confidence in elections.” Political Research Quarterly 62(3):507. Hammond, T. H. 1996. “Formal theory and the institutions of governance.” Governance-an International Journal of Policy and Administration 9(2):107–185. Hammond, T. H. and C. K. Butler. 2003. “Some complex answers to the simple question ’do institutions matter?’ Policy choice and policy change in presidential and parliamentary systems.” Journal of Theoretical Politics 15(2):145–200. Hammond, T. H. and J. H. Knott. 1996. “Who controls the bureaucracy?: Presidential power, congressional dominance, legal constraints, and bureaucratic autonomy in a model of multi-institutional policy-making.” Journal of Law Economics & Organization 12(1):119– 166. Hartlyn, J., J. McCoy and T. M. Mustillo. 2008. “Electoral governance matters - Explaining the quality of elections in contemporary Latin America.” Comparative Political Studies 41:73– 98. 323 Hearn, J. 2000. “Aiding democracy? Donors and civil society in South Africa.” Third World Quarterly 21(5):815–830. Helmke, G. and S. Levitsky. 2006. Informal institutions and democracy : lessons from Latin America. Baltimore: Johns Hopkins University Press. Herskovits, J. 2007. “Nigeria’s rigged democracy.” Foreign Affairs 86(4):115–+. Hirschman, Albert O. 1970. Exit, voice, and loyalty: Responses to decline in firms, organizations, and states. Vol. 25 Cambridge, Mass.: Harvard University Press. HRW. 2011. “Nigeria: Post-Election Violence Killed 800.” Human Rights Watch Report. Huber, J. D. and C. R. Shipan. 2006. “Politics, delegation, and bureaucracy.” The Handbook for Political Economy. Huber, J. D. and C. R. Shipan. 2002. Deliberate discretion: the institutional foundations of bureaucratic autonomy. Cambridge studies in comparative politics. New York: Cambridge University Press. Huber, J. D. and N. McCarty. 2004. “Bureaucratic Capacity, Delegation, and Political Reform.” The American Political Science Review 98(3):481–494. Huntington, S. P. 1991. The third wave: Democratization in the late twentieth century. University of Oklahoma Press. Huntington, Samuel P. 1968. Political order in changing societies. New Haven,: Yale University Press. Hyde, S. D. 2011. “Catch Us If You Can: Election Monitoring and International Norm Diffusion.” American Journal of Political Science. Ibrahim, J. 2007. Nigeria’s 2007 elections: the fitful path to democratic citizenship. United States Institute of Peace. 324 Ibrahim, J. and S. G. Egwu. 2005. Nigeria elections 2007 : defending the people’s mandate. Abuja: ICG. 2004. “Zimbabwe Another Election Chance” Africa Report: N 89. ICG. 2005. “Post-Election Zimbabwe: What Next?” Africa Report: N 93.” ICG. 2007. “Nigeria: Failed elections, Failing State.” Africa Report: N 126 ICG. 2011. “Lessons from Nigeria’s 2011 Elections.” Africa Briefing: N 81. IEC. 2009. “Electoral Commission of South Africa: 2009 Elections Report. James, T. S. 2012. Elite Statecraft and Election Administration: Bending the Rules of the Game? Palgrave Macmillan. JDPC. 2007. “Report of the Observation Of The 2007 General Elections In Nigeria.”. Jeffries, R. 1998. “The Ghanaian elections of 1996: towards the consolidation of democracy?” African Affairs 97(387):189. Jeffries, R. and C. Thomas. 1993. “The Ghanaian elections of 1992.” African Affairs 92(368):331. Jinadu, L. A. 2011. Ghana. Johannesburg: Open Society Foundations. Johnson, R. 2011. Preventing Electoral Fraud in Zimbabwe. South African Institute on Race Relations. Joseph, R. A. 1987. Democracy and prebendal politics in Nigeria: the rise and fall of the Second Republic. Cambridge; New York: Cambridge University Press. Kadima, D. and S. Booysen. 2009. Compendium of elections in Southern Africa, 1989- 2009: 20 years of multiparty democracy. Electoral Institute of South Africa. 325 Keefer, P. and D. Stasavage. 2003. “The limits of delegation: Veto players, Central Bank independence, and the credibility of monetary policy.” American Political Science Review 97(3):407–423. Kelley, J. and K. Kolev. 2010. “Election Quality and International Observation 1975-2004: Two New Datasets.”. Kelley, J. 2012. Monitoring democracy: When international election observation works, and why it often fails. Princeton University Press. Kelly, N. 2007. “The independence of electoral management bodies: The Australian experience.” Political Science 59(2):17–32. Kew, D. 1999. “ ‘Democrazy: Dem Go Craze, O’: Monitoring the 1999 Nigerian Elections.” Issue: A Journal of Opinion pp. 29–33. Kuenzi, M. and G. Lambright. 2005. “Party systems and democratic consolidation in Africa’s electoral regimes.” Party Politics 11(4):423–446. Kuntz, P. and M. Thompson. 2009. “More than just the final straw: Stolen elections as revolutionary triggers.” Comparative Politics pp. 253–272. Laakso, L. 2002. “The politics of international election observation: the case of Zimbabwe in 2000.” Journal of Modern African Studies 40(3):437–464. Lassen, D. 2005. “The effect of information on voter turnout: Evidence from a natural experiment.” American Journal of Political Science 49(1):103–118. Lehoucq, F. 2003. “Electoral Fraud: Causes, Types and Consequences.” Annual Review of Political Science 6:23. Lehoucq, F, and J. Molina. 2002. Stuffing the ballot box: fraud, electoral reform, and democratization in Costa Rica. Cambridge studies in comparative politics. New York: Cambridge University Press. 326 Leonard, D. K. and F. Odhiambo Owour. 2009. “The Political and Institutional Context of the 2007 Kenyan Elections and Reforms Needed for the Future.” Journal of African Elections 8(1):71–107. LeVan, A. C., T. Pitso and B. Adebo. 2004. “Elections in Nigeria: Is the Third Time a Charm?” Journal of African Elections 2(2):30–47. LeVan, A and P. Ukata. 2012. “Countries at the Crossroads 2012–Nigeria.” Freedom House . Levitsky, S. and L. Way. 2010. Competitive authoritarianism: hybrid regimes after the Cold War. Cambridge University Press. Lewis, P. M. 2003. “Elections in a Fragile Regime.” Journal of Democracy 14(3):131–144. Lewis, P. M. 2011. “Nigeria Votes: More Openness, More Conflict.” Journal of Democracy 22(4):60–74. Lieberman, E. S. 2005. “Nested analysis as a mixed-method strategy for comparative research.” American Political Science Review 99(3): 435–452. Lijphart, A. and C. Waisman. 1996. Institutional design in new democracies : Eastern Europe and Latin America. Latin America in global perspective. Boulder, Colo.: Westview Press. Lindberg, S. I. 2005. “Consequences of electoral systems in Africa: a preliminary inquiry.” Electoral Studies 24(1):41–64. Lindberg, S. I. 2006a. Democracy and elections in Africa. Baltimore: Johns Hopkins University Press. Lindberg, S. I. 2006b. Do Opposition Parties Boycott Elections? In Electoral authoritarianism : the dynamics of unfree competition, ed. Andreas Schedler. Boulder, Colo.: Lynne Rienner Publishers, Inc. Lodge, T., D. Kadima and D. Pottie. 2002. Compendium of elections in Southern Africa. Electoral Institute of Southern Africa. 327 Lopez-Pintor, R. 2000. Electoral Management Bodies as Institutions of Governance. New York: United Nations Development Program. Lopez-Pintor, R. and J. Fisher. 2006. Getting to the core: A Global survey on the cost of registration and elections. International Foundation of Electoral Systems. Lyons, T. 1997. “Ghana’s encouraging elections .1. A major step forward.” Journal of Democracy 8(2):65–77. Magaloni, B. 2006a. “The Game of Electoral Fraud and the Ousting of Authoritarian Rule.” American Journal of Political Science 54(3): 751–765. Magaloni, B. 2006b. Voting for autocracy: Hegemonic party survival and its demise in Mexico. Cambridge University Press Cambridge. Makulilo, A. B. 2009. “Independent Electoral Commission in Tanzania: A False Debate?” Representation 45(4): 435–453. Makumbe, J. 2002. “Zimbabwe’s hijacked election.” Journal of Democracy 13(4): 87–101. Massicotte, L. 2005. “Judging elections by their outcome?” Representation 41(3):180–188. Massicotte, L., A. Blais and A. Yoshinaka. 2004. Establishing the rules of the game: election laws in democracies. Toronto: University of Toronto Press. Masunungure, E. 2009. “Defying the winds of change: Zimbabwe’s 2008 elections.” Mattes, R. and C. Shenga. 2013. Uncritical Citizenship: Mozambicans in Comparative Perspective. In Voting and Democratic Citizenship in Africa, ed. Michael Bratton. Lynne Rienner Publishers chapter 9. Matyszak, D. 2009. Hear no evil, see no evil, speak no evil: a critique of the Zimbabwe Electoral Commission report on the 2008 general elections. Research and Advocacy Unit. McFaul, M. 2002. “The fourth wave of democracy and dictatorship - Noncooperative 328 transitions in the postcommunist world.” World Politics 54(2):212–+. Migdal, J. 1988. Strong societies and weak states: state-society relations and state capabilities in the Third World. Princeton, N.J.: Princeton University Press. Moehler, D. C. 2009. “Critical Citizens and Submissive Subjects: Election Losers and Winners in Africa.” British Journal of Political Science 39(02): 345–366. Moser, P. 1999. “Checks and balances, and the supply of central bank independence.” European Economic Review 43(8): 1569–1593. Mozaffar, S. 2002. “Patterns of electoral governance in Africa’s emerging democracies.” International Political Science Review 23(1):85–101. Mozaffar, S. and A. Schedler. 2002. “The comparative study of electoral governance Introduction.” International Political Science Review 23(1):5–27. NDI. 2007. “Statement of the National Democratic Institute (NDI) International Election Observer Delegation To Nigeria’s April 21 Presidential and National Assembly Elections.” NDI. 2012. “Statement of the National Democratic Institute (NDI) International Election Observer Delegation To Nigeria’s April Presidential and National Assembly Elections.” Nijzink, L., S. Mozaffar and E. Azevedo. 2006. “Parliaments and the enhancement of democracy on the African continent: An analysis of institutional capacity and public perceptions.” The Journal of Legislative Studies 12(3-4): 311–335. Norris, P. 2011. “Cultural Explanations of Electoral Reform: A Policy Cycle Model.” West European Politics 34(3):531–550. Norris, P. 2004. Electoral engineering : voting rules and political behavior. Cambridge, UK ; New York: Cambridge University Press. Norris, P. 2012. “Are there global norms and universal standards of electoral integrity and malpractice? Comparing public and expert perceptions.”. 329 North, D. 1990. Institutions, institutional change, and economic performance: The Political economy of institutions and decisions. Cambridge; New York: Cambridge University Press. Nwankwo, C. 1999. “Monitoring Nigeria’s Elections.” Journal of Democracy 10:156–165. Obi, C. I. 2008. “International election observer missions and the promotion of democracy: Some lessons from Nigeria’s 2007 elections.” Politikon 35(1):69–86. Ogbeidi, M. M. 2010. “A culture of failed elections: revisiting democratic elections in Nigeria, 1959-2003.” Historia Actual Online (21):43. Olsen, J. P. and J. G. March. 1984. “The new institutionalism: Organizational factors in political life.” American Political Science Review 78(3-4):734–749. Oquaye, M. 1995. 94(375): 259. “The Ghanaian Elections of 1992–A Dissenting View.” African Affairs Ostrom, E. 1990. Governing the commons : the evolution of institutions for collective action. Political economy of institutions and decisions. ed. Cambridge [England] ; New York: Cambridge University Press. Padmanabhan, V. 2002. “Democracy’s baby blocks: South Africa’s electoral commissions.” New York University Law Review 77(4):1157–1194. Pastor, R. A. 1999. “The Role of Electoral Administration in Democratic Transitions: Implications for Policy and Research.” Democratization 6(4): 28. Pierson, P. 2000a. “The limits of design: Explaining institutional origins and change.” Governance: An International Journal of Policy and Administration 13(4):475–499. Pierson, P. 2000b. “Increasing Returns, Path Dependence, and the Study of Politics.” The American Political Science Review 94(2): 251–267. Piper, L. 2005. South Africa’s 2004 election: the quest for democratic consolidation. EISA. Ploch, L. 2008. “Zimbabwe: 2008 Elections and Implications for US Policy.” 330 Posner, D. N. and D. J. Young. 2007. “The institutionalization of political power in Africa.” Journal of Democracy 18(3): 126–140. Pottie, D. 2003. “The presidential elections in Zimbabwe, March 2002.” Electoral Studies 22(3): 516–523. Pottie, David. 2002. “Parliamentary elections in Zimbabwe, 2000.” Electoral Studies 21(3):485–492. Powell, G. B. 2004. “The chain of responsiveness.” Journal of Democracy 15:91–105. Powell, G Bingham. 2000. Elections as instruments of democracy: Majoritarian and proportional visions. Yale University Press. Przeworski, A. 2000. Democracy and development: Political institutions and well-being in the world, 1950-1990. Cambridge University Press. Przeworski, A. 2004. “Institutions matter?” Government and Opposition 39(4): 527–540. Raftopoulos, B. and S. Eppel. 2008. “Desperately Seeking Sanity: What Prospects for a New Beginning in Zimbabwe?” Journal of Eastern African Studies 2(3): 369–400. Rakner, L. and N. van de Walle. 2009. “Opposition weakness in Africa.” Journal of Democracy 20 (3): 108–121. Rauch, J. E. and P. B. Evans. 2000. “Bureaucratic structure and bureaucratic performance in less developed countries.” Journal of Public Economics 75(1): 49–71. Raudenbush, S. W. and A. S. Bryk. 2002. Hierarchical linear models: Applications and data analysis methods. Sage Publications. Rawlence, B. and C. Albin-Lackey. 2007. “Briefing: Nigeria’s 2007 general elections: Democracy in retreat.” African Affairs 106(424):497–506. Rosas, G. 2010. “Trust in elections and the institutional design of electoral authorities: 331 Evidence from Latin America.” Electoral Studies 29(1):74–90. Rose, R. and W. Mishler. 2009. “How Do Electors Respond to an” Unfair” Election? The Experience of Russians.” Post-Soviet Affairs 25(2): 118–136. SADC. 2004. “SADC Principles and Guidelines Governing Democratic Elections.” Sartori, G. 1970. “Concept misformation in comparative politics.” The American Political Science Review 64 (4)1033–1053. Schaffer, F. 2008. The hidden costs of clean election reform. Ithaca: Cornell University Press. Schedler, A. 2002a. “The menu of manipulation.” Journal of Democracy 13(2):36–50. Schedler, A. 2002b. “The nested game of democratization by elections.” International Political Science Review 23(1):103–122. Schedler, A., L. Diamond and M. Plattner. 1999. The self-restraining state: power and accountability in new democracies. Boulder, Colo.: Lynne Rienner Publishers. Seawright, J. and J. Gerring. 2008. “Case selection techniques in case study research: A menu of qualitative and quantitative options.” Political Research Quarterly 61(2):294. Sheplse, K. 1989. “Studying Institutions. Some Lessons from the Rational Choice Approach.” Journal of Theoretical Politics 1:131–147. Sives, A. 2009. “Electoral Reform and Good Governance: The Case of Jamaica.” Commonwealth & Comparative Politics 47(2):174–193. South Africa Electoral Commission Act 1996. Steenbergen, M. R. and B. S. Jones. 2002. “Modeling multilevel data structures.” American Journal of Political Science pp. 218–237. 332 Steinmo, Sven, Kathleen Ann Thelen and Frank Longstreth. 1992. Structuring politics: historical institutionalism in comparative analysis. Cambridge studies in comparative politics. Cambridge [England] ; New York: Cambridge University Press. Suberu, R. T. 2007. “Nigeria’s muddled elections.” Journal of Democracy 18(4):95–110. Svensson, P. and J. Elklit. 1997. “What Makes Elections Free and Fair?” Journal of Democracy 8(3):32–46. Tarrow, Sidney. 1995. “Review: Bridging the Quantitative-Qualitative Divide in Political Science.” The American Political Science Review 89(2):471–474. TCC/NDI. 1999. “Statement of the National Democratic Institute (NDI) and The Carter Center (TCI) International Election Observer Delegation To Nigeria’s 1999 Elections.” Thelen, K. 1999. “Historical institutionalism in comparative politics.” Annual Review of Political Science 2:369–404. Thelen, K. and J. Mahoney. 2009. Explaining institutional change. Cambridge University Press Textbooks. Ting, M. 2005. “Organizational capacity.” Typescript, Department of Political Science, Columbia University. TMG. 2003. “Do the Votes Count: Final Report of the 2003 General Elections in Nigeria.” TMG. 2007. “An election programmed to fail: final report of the April 2007 general election in Nigeria.”. Tomz, M. and R. Van Houweling. 2003. “How Does Voting Equipment Affect the Racial Gap in Voided Ballots?” American Journal of Political Science 47(1):46–60. Trebilcock, M. and P. Chitalkar. 2009. “From Nominal to Substantive Democracy: The Role and Design of Election Management Bodies.” The Law and Development Review. Tsai, K. S. 2006. “Adaptive informal institutions and endogenous institutional change in 333 China.” World Politics 59(1):116. van de Walle, N. 2003. “Presidentialism and Clientelism in Africa’s emerging party system.” Journal of Modern African Studies 41(2):25. van de Walle, N. 2002. “Africa’s range of regimes.” Journal of Democracy 13(2):66–80. Vengroff, R. and M. Ndiaye. 1998. “The impact of electoral reform at the local level in Africa: The case of Senegal’s 1996 local elections.” Electoral Studies 17(4):463–482. Wall, Alan, International Institute for Democracy and Electoral Assistance. 2006. Electoral management design : the International IDEA handbook. Stockholm: International IDEA. Weaver, R. Kent and Bert A. Rockman. 1993. Do institutions matter?: Government capabilities in the United States and abroad. Washington, D.C.: The Brookings Institution. Wright, J. 2009. “How Foreign Aid Can Foster Democratization in Authoritarian Regimes.” American Journal of Political Science 53(3):552–571. Yar’Adua, Umaru. 2007. “Inauguration Speech [Transcript].” ZEC. 2008. “ Zimbabwe Electoral Commission Report on the 2008 General Elections.” ZESN. 2005. “Zimbabwe Election Support Network (ZESN) Report on the Zimbabwe’s 2005 general election, 31 April 2005.” “ ZESN. 2008. “Zimbabwe Election Support Network (ZESN): Report on the Zimbabwe 29 March 2008 harmonized elections and 27 June presidential run-off. 334