拉丁美洲選舉廢票之研究 (1916-2018) - 政大學術集成

全文

(1)國立政治大學亞太研究英語碩士學位學程 International Master’s Program in Asia- Pacific Studies College of Social Sciences National Chengchi University. 碩士論文. 治政 Thesis Master’s. 立. 大. ‧ 國. 學 ‧. 拉丁美洲選舉廢票之研究 (1916-2018). Nat. n. al. er. io. sit. y. Explaining Invalid Voting in Latin America (1916-2018). Ch. engchi. i n U. v. Student: Carolina Lui Lam Postigo 林伊娜 Advisor: Yen-Pin Su, Ph. D. 蘇彥斌教授. 中華民國 109 年 7 月 July 2020. 1 DOI:10.6814/NCCU202100292.

(2) Abstract What explains the variation in invalid votes across different countries? This thesis extends our knowledge about factors that impact voters’ propensity to cast invalid ballots by testing economic, sociological, and institutional explanations. Unlike previous studies, this thesis contributes to the literature by providing systemic empirical analyses that cover samples of elections across a century. Drawing on evidence from presidential elections and lower chamber elections in Latin America between 1916 and 2018, this study demonstrates four findings. First, a country with an enforced compulsory voting system tends to have higher levels of invalid vote for lower chamber elections than presidential elections. Second, while the level of regime corruption has a positive effect on increasing invalid votes, such an effect is stronger for lower chamber elections than presidential elections. Third, concurrent elections tend to increase invalid vote for presidential. 政治大. elections but not for lower chamber elections. Fourth, a higher level of economic development tends to increase invalid vote for lower chamber elections but not for presidential elections.. 立. Nat. 摘要. y. ‧ 國. ‧. corruption. 學. Keywords Latin America, presidential elections, legislative elections, invalid voting, compulsory voting,. sit. 為什麼有些國家的選舉會出現比較多的廢票、有些卻較少？本論文試圖從經濟、社會與制. n. er. io. 度的角度對於這個問題進行探討。與既有文獻不同的是，本論文的貢獻在於其實證分析包 al 含超過一世紀的選舉資料。本研究以拉丁美洲各國總統選舉與下議院國會選舉（1916iv n C 1918）資料為基礎，得出四個研究發現。一、對於確實執行強制性投票制度的國家而言， hengchi U 其國會選舉的廢票率會比總統選舉的廢票率來得高；二、國家體制的貪腐程度會增加一國國會選舉與總統選舉的廢票率，但這個增強效果對於國會選舉的影響較大；三、當總統選舉與國會選舉同時舉行時，總統選舉的廢票率會比國會選舉的廢票率來得高；四、對於經濟發展程度較高的國家而言，其國會選舉的廢票率會比總統選舉的廢票率來得高。關鍵字拉丁美洲；總統選舉；國會選舉；廢票；強制性投票；貪腐. 2 DOI:10.6814/NCCU202100292.

(3) Resumen ¿Qué explica esta variación del voto inválido entre los países? Utilizando los enfoques económicos, sociológico e institucional, esta tesis extiende nuestra comprensión sobre los factores que afectan la decisión del electorado y su opción de invalidar el voto. Al contrario de los estudios anteriores, esta tesis proporciona un análisis empírico-sistemático sobre la base de una muestra electoral de alrededor de un siglo. Así, sobre la base de las evidencias encontradas en las elecciones presidenciales y legislativas en América Latina, entre 1916 y 2018, este estudio sostiene cuatro resultados. Primero, un país con un sistema de votación obligatorio forzado tiende a tener niveles más altos de votos inválidos para las elecciones legislativas que para las elecciones presidenciales. Segundo, si bien el nivel de corrupción del régimen tiene un efecto positivo en el aumento de los votos inválidos, ese efecto es mayor para las elecciones legislativas que en las presidenciales. En tercer lugar, las elecciones concurrentes tienden a aumentar el voto inválido en las elecciones presidenciales, pero no para las elecciones legislativas. En cuarto lugar, un mayor nivel de desarrollo económico tiende a incrementar el voto inválido para las elecciones legislativas y no, para las presidenciales.. 立. ‧ 國. 學. Palabras claves. 政治大. sit. io. ‫ملخص الدراسة‬. er. Nat. y. ‧. América Latina, elecciones presidenciales, elecciones legislativas, voto inválido, voto obligatorio, corrupción. n. a l‫الباطلة عبر البلدان المختلفة؟ ت سعى هذه‬ ‫األطروحة للتعرف على العوامل التي تؤثر على‬ i v ‫ما الذي يفسر االختالف في األصوات‬ n C ‫ على عكس الدراسات‬.‫االقتصادية واالجتماعية والمؤسسية‬ ‫اختبار‬ h e‫التفسيرات‬ i U‫ عن طريق‬،‫قيام الناخبين باإلدالء بأصوات باطلة‬ h n c g .‫ تسااااااهو هذه األطروحة في األدبيات من خالل توفير تبليالت تجريبية منعجية تمني عينات من االرتخابات عبر قرن‬،‫الساااااابسة‬ .2018 ‫ و‬1916 ‫وباالعتماد على األدلة من االرتخابات الرئاساااية وارتخابات المجالس التشاااريعية في أمريكا الالتينية بين عامي‬ ‫ تبصاااال الدولة التي لديعا رظام تصااااويز مل امي قسااااري على مسااااتويات أعلى من‬:‫ أوالا‬، ‫كما توضااااه هذه الدراسااااة أرب رتائ‬ ‫ في حين أن مساااتو فسااااد النظام لأ تأثير‬:‫ ثارياا‬.‫التصاااويز الباطل الرتخابات المجالس التشاااريعية مساررةا باالرتخابات الرئاساااية‬ :‫ ثالااا‬.‫ ف ن تأثير هذا يكون أقو في ارتخابات المجالس التشااريعية عن االرتخابات الرئاسااية‬،‫ميجابي على زيادة األصااوات الباطلة‬ .‫تؤدي االرتخابات المت امنة ملى زيادة األصاااوات ير الصاااالبة لالرتخابات الرئاساااية ولكن ليس الرتخابات المجالس التشاااريعية‬ ‫ يؤدي المستو األعلى من التنمية االقتصادية ملى زيادة األصوات الباطلة الرتخابات المجالس التشريعية وليس االرتخابات‬:‫رابعاا‬ .‫الرئاسية‬ ‫الكلمات الدالة‬ .‫ فساد‬، ‫ اقتراع مل امي‬، ‫ تصويز باطل‬، ‫ ارتخابات تشريعية‬، ‫ ارتخابات رئاسية‬، ‫أمريكا الالتينية‬. 3 DOI:10.6814/NCCU202100292.

(4) Table of Contents. Abstract ......................................................................................................................................2 Table of Contents ........................................................................................................................4 List of Figures ..............................................................................................................................6 List of Tables ...............................................................................................................................7 Chapter One: Introduction ..........................................................................................................8 1.1 Puzzle ............................................................................................................................8. 政治大 Chapter Two: Theoretical Perspective 立.......................................................................................18 1.2 Why Study Invalid votes? .............................................................................................15. ‧ 國. 學. 2.1 Institutional framework ................................................................................................18 2.2 Socioeconomic characteristics ......................................................................................20. ‧. 2.3 Political context ............................................................................................................21 Chapter Three: Research Design................................................................................................24. y. Nat. sit. 3.1 Large- N Tests ...............................................................................................................24. er. io. 3.1.1 Dependent Variable ..................................................................................................24. n. al 3.1.2 Independent Variables ..............................................................................................26 iv C. n. hen hi U 3.1.3 Control Variables .......................................................................................................27 gc. 3.2 Methods and Estimation Techniques ............................................................................31 Chapter Four: Quantitative Analysis ..........................................................................................32 4.1 Invalid Voting in Latin America .....................................................................................32 4.2 Empirical Results ..........................................................................................................37 4.3 Robustness Check.........................................................................................................43 Chapter Five: Qualitative Analysis .............................................................................................46 5.1 Invalid Vote Campaigns for Presidential Elections................................................................47 5.1.1 Argentina...................................................................................................................47 5.1.2 Bolivia........................................................................................................................49 4 DOI:10.6814/NCCU202100292.

(5) 5.1.3 Brazil .........................................................................................................................49 5.1.4 Chile ..........................................................................................................................52 5.1.5 Colombia ...................................................................................................................53 5.1.6 Costa Rica ..................................................................................................................54 5.1.7 Ecuador .....................................................................................................................55 5.1.8 Guatemala .................................................................................................................56 5.1.9 Mexico ......................................................................................................................56 5.1.10 Peru .........................................................................................................................56 5.2. Lower Chamber elections ............................................................................................59 5.2.1. Argentina..................................................................................................................59 5.2.2. Colombia ..................................................................................................................60. 政治大. 5.2.3. Ecuador ....................................................................................................................61. 立. 5.2.4. El Salvador ................................................................................................................61. ‧ 國. 學. 5.2.5. Mexico .....................................................................................................................61 5.3 Comparative Analyses ..................................................................................................63. ‧. Chapter Six: Conclusion .............................................................................................................71. sit. y. Nat. 6.1 Research Summary .......................................................................................................71 6.2 Policy Implication .........................................................................................................73. er. io. n. References ................................................................................................................................75 a v. i l C n hengchi U. 5 DOI:10.6814/NCCU202100292.

(6) List of Figures. Figure 1: Invalid vote by World Region (1980-2019) ..................................................................10 Figure 2: Percent Invalid vote in Presidential and Lower Chamber elections among Latin American countries (1916-2018) ...............................................................................................................11 Figure 3: Invalid vote rates in presidential elections by Latin American countries Pre vs Post the Third Wave of Democratization .................................................................................................13. 治政 the Third Wave of Democratization...........................................................................................14 大立 Figure 4: Invalid vote rates in Lower Chamber elections by Latin American countries Pre vs Post. ‧ 國. 學. Figure 5: Percentage of Invalid Vote in presidential elections in Latin America ..........................33 Figure 6: Percentage of Invalid Vote in Lower Chamber elections in Latin America ...................35. ‧. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. 6 DOI:10.6814/NCCU202100292.

(7) List of Tables. Table 1: Observations Included for the Statistical Analyses .......................................................24 Table 2: List of Variables............................................................................................................28 Table 3. Summary Statistics of the Variables Used in the Empirical Analyses for presidential election .....................................................................................................................................30 Table 4. Summary Statistics of the Variables Used in the Empirical Analyses for lower chamber election .....................................................................................................................................30 Table 5. Results for Model of Invalid Voting in Latin America Presidential elections ..................38. 政治大. Table 6. Results for Model of Invalid Voting in Latin America lower chamber elections .............40. 立. Table 7. Consistency of Results for Presidential Elections and Lower Chamber Elections ...........42. ‧ 國. 學. Table 8. Robustness Checks.......................................................................................................44. ‧. Table 9. Invalid Vote Campaigns in Latin America Presidential Elections (1999-2019) ...............64. sit. y. Nat. Table 10. Invalid Vote Campaigns in Latin America Lower Chamber Elections (2001-2018) .......68. n. al. er. io. Table 11. T-Tests for Anti-Corruption Invalid Vote Campaigns and Levels of Invalid Vote ..........69. Ch. engchi. i n U. v. 7 DOI:10.6814/NCCU202100292.

(8) Chapter One: Introduction. 1.1 Puzzle Elections are crucial for the proper functioning of the country’s political, social, and economic development, especially for democratic countries (Powell 1982). Elections allow citizens to accept or reject their representatives and their policy platforms (OSCE 2013), but it is common to observe null and blank ballots in every election. Those ballots are considered as invalid votes, and they reflect no support for parties or candidates. The presence of invalid voting. 政治大 2018; Power and Garand 2007; Superti 2013; Uggla 2008). While some studies contend that 立 invalid voting results from the absence of voters’ political skills or information about the elections. has gained scholarly attentions (e.g., Cohen 2018; Lysek et al. 2019; Martinez I Coma and Werner. ‧ 國. 學. (e.g., Pachon et al. 2017; Pierzgalski et al. 2019), other studies suggest that invalid voting is an expression of voter discontent or apathy (e.g., Cohen 2018; Power and Garand 2007; Uggla 2008).. ‧. Chilean Senator Navarro B. Alejandro argues that "The blank vote is a rejection against politicians. Nat. sit. y. and their programs, symbol of democracy, which gains adherents every day and is considered by. er. io. many as an exemplary protest and as the best way to reject corruption and abuse of the oligarchic. n. system of political parties" (Navarro a 2010). Similarly, Malamud (2018, v 5) mentioned, “The ‘anger. l. i. vote’ is a cathartic act which responseC toh the grave andiwidespread unrest and discontent with Un national institutions and representatives.”. engch. Using invalid voting as a form of political protest, influential elites or social groups might encourage the supporters to spoil their vote during a democratic crisis and repetitive political disenchantment. For instance, during the Argentine Constitutional Assembly Election of 1957, Domingo Peron called on voters to cast blank ballots (Baeza 2016). Eventually, invalid votes topped with 25% of the total votes, considering it as a political signal of the Peronist Resistance (Baeza 2016). In Argentina's 2001 legislative election, a strong campaign encouraged people to cast the so-called "voto bronca" (anger vote). It was an action for "the anger of citizens with the political class, which many accuse of corrupt and inoperative to get the country out of the severe 8 DOI:10.6814/NCCU202100292.

(9) economic recession" (Sabanes 2001). In this manner, the "voto bronca" became a "democratic option" for argentine voters (EL País 2001). During the second round of Argentina's 2015 presidential election, Nicolás del Caño (candidate for the Left and the Workers' Front) called to cast a blank ballot as an opposition to the political agenda of Mauricio Macri and Daniel Scioli (La Izquierda Diario 2015). In Mexico, during the Lower Chamber election in 2009, campaigns for null votes through social networks such as #VeVotaAnula or "EllosNoNosRepresentan" ("They do not represent us") were regarded as flag of protests against the high rates of crimes, economic crises, and impunity in cases of corruption (Cisneros 2012). In Peru, during the second round of Presidential elections. 政治大. in 2001, Alvaro Mario Vargas Llosa and Jaime Bayly promoted the blank vote, arguing that "We are going to register a third candidate. That is the only clean candidate we can trust; the blank. 立. vote" (La Nación 2001). In the 2017 Ecuadorian presidential election, Jorge Cedeño López, the. ‧ 國. 學. leader of the "Organization, Production and Development" movement, led a strong null vote campaign (El Universo 2017). In the 2018 Colombian presidential election, two influential. ‧. politicians, Sergio Fajardo y Humberto de la Calle, stated their support for the blank vote,. sit. y. Nat. declaring that neither the right-wing candidate Ivan Duque nor the left-wing candidate Gustavo Petro was convincing them (Gonzales 2018).. er. io. n. a l across different countriesi vand periods. Among different Invalid voting varies widely Un. C. i in Presidential and Lower Chamber eng continents, Latin America has the highest h invalid voting c h rates elections in the world. On average, invalid ballots in Latin America represented more than 5 percent of all votes cast in Presidential elections and nearly reached to 9 percent of all ballots observed for Lower Chamber elections. In contrast, in regions such as Europe and Asia, rates of invalid voting are less than 3 percent on average. Hence, the high invalid votes in Latin America poses an interesting empirical puzzle.. 9 DOI:10.6814/NCCU202100292.

(10) Figure 1: Invalid vote by World Region (1980-2019) 10 9 8. % Invalid vote. 7 6 5 4. 3 2 1 0. 立. Latin America. Asia. 政治大 Africa. Presidential. Europe. Oceania. Lower Chamber. ‧ 國. 學. Source: International Institute for Democracy and Electoral Assistance (International IDEA 2019). ‧. y. Nat. Amid a climate of political instability and citizens' support for democracy decreasing. io. sit. (Corporación Latinobarómetro 2018), Latin America faces a wave of significant increasing of. n. a. er. invalid votes of the last four decades. As Figure 2 shows, the invalid voting rates have been. v. i presidential elections, except l elections than that fornthe generally higher for the lower chamber. Ch. U. e inn Mexico, g c h i Paraguay, and Uruguay are the lowest Bolivia and Panama. The invalid voting rates. among all Latin American countries, while the invalid voting rates in Brazil, Ecuador, and Peru are the highest. The empirical data suggest that focusing on only a few explanatory variables might not be sufficient to explain the variation of invalid voting in different countries. For instance, Ecuador, Peru and Brazil, where compulsory voting is strictly enforced by a system of sanctions displayed the highest percentages of invalid votes. However, using similar electoral regulation, Uruguay had the lowest level of invalid votes. In countries where voting is compulsory but without penalties, invalid voting rates were high in Guatemala and El Salvador but not in Costa Rica and. 10 DOI:10.6814/NCCU202100292.

(11) Paraguay. In Chile (post-2012), Mexico, and Colombia, where voting is voluntary, the percentage of invalid votes is low for presidential elections only.. Figure 2: Percent Invalid vote in Presidential and Lower Chamber elections among Latin American countries (1916-2018) 30. 25. % Invalid vote. 20. 15. 立. 10. ‧ 國. ‧. 0. 學. 5. 政治大. n. er. io. sit. y. Nat. al. i CPresidential Parliament n hengchi U. v. Source: Author. Data are from Nohlen (2005a; 2005b) and International IDEA (2019).. Democratization process might matter for explaining the variation in invalid voting. As Mainwaring and Pérez- Liñán (2005) argue, the third wave of democratization has been one of the most extensive around the world, involving a high number of countries and lasted for longer than any past waves. The overall trend towards democratization started in different years for each Latin American country. The first countries to celebrate them were the Dominican Republic and Ecuador in 1978; Peru in 1980, Honduras in 1981, followed by Bolivia (1982), Argentina (1983), Uruguay, and Brazil (1985); Paraguay, Panama, and Chile in 1989, Nicaragua (1990), El 11 DOI:10.6814/NCCU202100292.

(12) Salvador (1994), Guatemala (1996). Finally, Mexico incorporated institutional mechanisms to enable the holding of competitive elections since 1997. Before the Third Wave of Democratization, Costa Rica reached its democratic regime in 1949, followed by Colombia and Venezuela in 1958 (Alcántara and Freidenberg 2006). As Power and Garand (2007) contend, elections held in more democratic countries should generate lower levels of invalid voting. Did the third wave democratization affect the variation of invalid votes in Latin America? Figure 3 compares the average rates of invalid voting in presidential elections in sixteen Latin American countries before and after the third wave democratization. In eight of the sixteen Latin countries that held presidential elections such as. 政治大. Argentina, Bolivia, Dominican Republic, El Salvador, Guatemala, Mexico, Nicaragua, and Panama, the variation of invalid votes was lower after the third wave. Nevertheless, contrary to what. 立. Power and Garand (2007) have theorized, in countries such as Brazil, Chile, Colombia, Ecuador,. ‧ 國. 學. Honduras, Paraguay, Peru, and Uruguay, the average rates of invalid voting were higher after the third wave. Moreover, Figure 4 compares the average rates of invalid voting in lower chamber. ‧. elections in thirteen Latin American countries before and after the third wave democratization.. sit. y. Nat. Again, contrary to Power and Garand (2007), the invalid voting rates are higher in the post-third. io. n. al. er. wave period for most Latin American countries.. Ch. engchi. i n U. v. 12 DOI:10.6814/NCCU202100292.

(13) 立. 政治大. 學 After 3rd WD. y. ‧. Nat. Before 3rd WD. io. sit. 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0. ‧ 國. % INVALID VOTE. Figure 3: Invalid vote rates in presidential elections by Latin American countries Pre vs Post the Third Wave of Democratization. er. Notes: The availability of information limits the range of years of each country: Argentina (1916 to 2015); Bolivia. al. n. iv n C - 2018); Nicaragua (1984 - 2016); Panama (1948 h - 2014); i U2018); Chile (1938-2017), Colombia (1946 e n Brazil g c h(1945(1951- 2014); Dominican Republic (1962 - 2016); El Salvador (1984 - 2014); Guatemala (1958 - 2015); Mexico (1982. 2018); Ecuador (1968-2017); Honduras (1948-2017); Paraguay (1958-2018); Peru (1956-2016) and Uruguay (1984 2014). Likewise, due to the limited information, Costa Rica and Venezuela were not included in this Figure. Source: Author. Data are from Nohlen (2005a; 2005b) and International IDEA (2019).. 13 DOI:10.6814/NCCU202100292.

(14) Figure 4: Invalid vote rates in Lower Chamber elections by Latin American countries Pre vs Post the Third Wave of Democratization 25. 20. 15. 10. 5. 0. 立. 政治大. ‧ 國. 學. Before 3rd WD. After 3rd WD. ‧. y. Nat. Notes: The availability of information limits the range of years of each country: Argentina (1916 -2015); Bolivia (1956-. sit. 2014); Brazil (1950-2018); Colombia (1951-2018); Dominican Republic (1962-2016); Ecuador (1962-2017); El. al. er. io. Salvador (1985-2018); Guatemala (1959-2015); Mexico (1970-2018); Panama (1978-2014); Paraguay (1963-2018);. v. n. Uruguay (1971-2014) and Venezuela (1947-2005). Due to the limited information, Chile, Costa Rica, Honduras,. Ch. Nicaragua and Peru were not included in this Figure.. engchi. i n U. Source: Author. Data are from Nohlen (2005a; 2005b) and International IDEA (2019).. The growing invalid votes in Latin America elections has been a remarkable phenomenon. The above discussion shows that the level of Latin America’s invalid vote is higher than that of other regions, and there is a great variation of invalid voting rates within the region. Moreover, it seems that electoral systems and the democratization process are insufficient to explain the variation in invalid voting rates. Although, it is expected a different situation when this variation of invalid voting comes from the voter discontent or apathy against corruption committed by politicians or the incumbent government and its economic measures as it is seen in the context of mentioned invalid vote campaigns. 14 DOI:10.6814/NCCU202100292.

(15) Therefore, the invalid voting in Latin America poses an empirical puzzle. To address this puzzle, this thesis aims to systematically examine the factors that impact voters’ propensity to cast invalid votes. Specifically, I will test hypotheses regarding compulsory voting system, concurrent elections, level of economic performance, level of education, level of democracy, and corruption on invalid voting rates. Based on data of presidential elections and lower chamber elections in 18 Latin American countries from 1916 to 2018, I conduct mixed-method analyses to analyze the impacts of different factors. I conduct quantitative analyses to show the relative importance of each variable on invalid vote. I also conduct qualitative analyses to illustrate how invalid voting campaigns work in Latin America. In the aggregate, invalid voting responds predictably to certain features of institutional, socioeconomic, and political such as compulsory. 政治大 for the presidential and lower chamber 立 election. And while I find evidence that a country with an. voting system, concurrent elections, economic development, and corruption, in a different way,. ‧ 國. 學. enforced compulsory voting system and a high level of corruption affect more the increasing of invalid vote rates for lower chamber elections than presidential elections like it happened in. ‧. Argentina 2001 and Ecuador 2006 elections. Concurrent elections tend to increase invalid vote only for presidential elections; and, a higher level of economic development tends to increase. y. Nat. sit. invalid vote just for lower chamber elections. Overall, this study aims to fulfill the gap in the. n. al. er. io. literature and provide policy implications for voting behavior and democratization.. 1.2 Why Study Invalid votes?. Ch. engchi. i n U. v. Generally speaking, there are two types of invalid vote. The first type is a ballot left in unmarked, which is often called “blank votes.” The second type is a ballot that is marked in a way that makes it impossible to identify the voter’s choice, or a ballot that has been torn or defaced, which is often called “null votes” or “spoiled votes.” Following previous literature (Cohen 2018; Lysek et al. 2019; Power and Garand 2007; Uggla 2008), this thesis considers blank votes, null votes, and spoiled votes as invalid votes, and this thesis does not distinguish whether an invalid vote is cast intentionally or by error. Voting is arguable the most remarkable feature of a democratic regime by which voters' political preferences manifest in a single act. Hence, understanding the factors that affect the 15 DOI:10.6814/NCCU202100292.

(16) voter's choice to cast invalid votes is of crucial importance in the country's political life. Whether citizens cast invalid votes intentionally or accidentally, an extraordinarily high percentage of invalid votes might gravely undermine the well-functioning of the electoral system (Lysek et al. 2019) as well as democratic legitimacy (Lysek ep.al 2019; Power and Grand 2007). This issue is particularly crucial for new democracies as Latin America (Kouba and Lysek 2016). Moreover, although invalid votes do not count in favor of a particular political party or candidate, they are not innocuous in tight electoral contests. For instance, in a highly competitive election, a small portion of null or blank votes could make a big difference in the electoral results, such as enhance the probability of holding a run-off election (Cohen 2018).. 政治大 voting rates might represent that voters 立 are generally disenchanted with the candidates, political Another implication is related to the government performance. High levels of invalid. ‧ 國. 學. parties, or the incumbent government, which could lead to political instability. In this sense, invalid voting might not only serve as a mean of protest, but also a way to change the. ‧. configuration of the electoral scenario. Therefore, high levels of invalid voting rates could push the ruling party to coordinate with other political actors and improve the government. y. Nat. er. io. invalid voting than others.. sit. performance (Cohen 2016). In short, it is crucial to understand why some countries have higher. al. n. iv n C Why is it important to study invalid in Latin America? First, all Latin American h e nvote gchi U. countries regulate the electoral process through a special law which involves an explicit section. to those categories of votes often be regarded as invalid. In Argentina, it is regulated by National Electoral Code - Decree N° 2135 (article 101), Bolivia adopted the Electoral System Law- Law N° 026 (article 161) while Brazil enacted the Electoral Code- Law N° 4.737 (article 224). Chile regulates the electoral process through the Law of Voting and Popular indexes- Law N° 18.700 (article 71). In Colombia, the Electoral Code - Decree N° 2241 (article 137) besides, in Costa Rica, the Electoral Code - Law N° 8765 (article 194), Dominican Republic, the Electoral Law- Law N° 275-97 (article 129), Ecuador is the Electoral Law- Law N° 59-86 (article 59) and, in El Salvador is the Electoral Code - Decree N° 413 (article 207). In Guatemala, it is ruled by the Electoral and. 16 DOI:10.6814/NCCU202100292.

(17) Political Parties Law-Decree 1-85 (article 237), in Honduras, the Elections and Political Organizations Act - Decree N° 44-2004 (article 207). For Mexico, the General Law of Electoral Institutions and Procedures published was on January 27, 2017 (article 288), while in Nicaragua, it is regulated by the Electoral Law- Law N° 331 (article 125). Likewise, Panama has the Electoral Code (article 268), Paraguay, the Electoral Code Nº 834/96 (articles 225 and 226), Peru is the Electoral Law- Law Nº 26859 (article 286), Uruguay with the Electoral Law- Law Nº 7812 (article 106) and Venezuela adopted the Organic Law for Electoral Processes in 2012 (article 106). Second, as shown in the previous section, invalid voting is much higher in Latin America. 政治大 countries or regions (Aldashev and 立Mastrobuoni 2019; Hill and Young 2007; Martins 2017; than in other regions of the world. However, compared to studies of invalid votes in other. ‧ 國. 學. McAllister and Makkai 1993; Zulfikarpasic 2001), the issue of invalid votes in Latin America has received considerably less attention. While there are a number of single-country studies of invalid. ‧. votes for Latin America (Arbache et al. 2015; Cisneros 2012; Driscoll and Nelson 2014; Power and Roberts 1995; Sanchez 2016; Sanchez Berzain 2019), there are few comparative analyses across. y. Nat. sit. Latin American countries (e.g., Cohen 2018; Power and Garand 2007). Unlike those studies. er. io. mentioned above, this study is unique because it uses comprehensive electoral data in Latin. n. America to examine determinantsaofl invalid votes, and thus this i vthesis might facilitate a better. Un. C. e nofginvalid understanding of what explains different h levels c h i votes across time and space.. 17 DOI:10.6814/NCCU202100292.

(18) Chapter Two: Theoretical Perspective. To explain why some countries, have a higher level of invalid vote while others do not, previous studies have focused on three approaches: institutional framework, socioeconomic characteristics, and political context (Martinez I Coma and Werner 2018; Power and Garand 2007). Each approach focuses on a number of variables, which will be discussed as follows.. 2.1 Institutional framework Studies that focus on institutional factors tend to explain the occurrence of invalid votes. 政治大 difficult, desirable, or not to turn out to the polls (Power and Garand 2007). For example, 立 by examining the institutional design, legal and formal structures that make voting easy or. ‧ 國. 學. Pierzgalski et al. (2019) argue that the implementation of a complex or ambiguous ballot design might result in more invalid votes. Pachon et al. (2017) find that simplifying the ballot design. ‧. could explain the decrease of spoiled ballots in Colombia observed in 2007. Also, the introduction of electronic voting machines could influence the rates of ballot invalidation, as it is shown in. y. Nat. sit. Brazil (Nicolau 2015). Furthermore, the combination of high district magnitude with personalized. n. al. er. io. voting tends to increase the percentage of spoiled votes (Power and Garand 2007).. iv. n structures, and party system Relatedly, there are certain C electoral rules, ballot hengchi U. configurations that may favor the casting of invalid votes. Significantly, there are two elements of the institutional framework that have been suggested to affect the number of invalid votes: compulsory voting and whether presidential election and lower chamber election are held concurrently. Compulsory voting may respond to many plausible conceptions but it encloses a legal standpoint imposed by the national constitutions and electoral laws to increase voter turnout (Luiss 2008). Although compulsory voting law can certainly raise and equalize electoral turnout, it is questionable if forcing people to cast would be beneficial for democracy (Smith 2020). For instance, Gratschew (2002) finds that compulsory voting could encourage dissatisfied voters to 18 DOI:10.6814/NCCU202100292.

(19) cast invalid votes, even though the government contends that compulsory voting is necessary to fully perform civic duty. Studies of invalid voting find that there is a strong positive correlation between compulsory voting and the percentage of invalid votes (Blais 2006; Kouba and Lysek 2016; Martinez I Coma and Werner 2018; Power and Garand 2007; Power and Roberts 1995; Uggla 2008). Studying democratic Lower Chamber elections in 73 countries on five continents from 1970 to 2011, Martinez I Coma and Werner (2018) argued that both enforced and unenforced compulsory systems increase the level of invalid voting. Power and Garand’s (2007) analysis from 80 legislative elections held in 18 Latin American democracies shows that the combination of compulsory voting with sanctions,1 which are strongly enforced, increases the propensity to cast. 政治大. invalid ballots. Similarly, this assumption is a response from politically unsophisticated and. 立. uninterested voters as well as those disaffected with democracy who would rather not vote at. ‧ 國. 學. all and may invalidate their ballots as a protest signal (Engelen 2007; Singh 2015).. ‧. On the other hand, in a presidential system, voters enjoy a maximum discretion to express their preferences over political representation on the leadership of the executive and legislative. y. Nat. sit. branches (Carey 1998) through clear identification of their favorite candidate in their ballots.. er. io. Nevertheless, that fact may be altered by a factor in the electoral cycle, such as concurrent. n. a l 2009). Many studies have elections (Dettrey and Schwindt-Bayer i v shown that when presidential Un. C. e the election and legislative election are heldhon n gsame c h idate, voter turnout tends to be higher (Carreras 2017; Dettrey and Schwindt-Bayer 2009; Stockemer and Calca 2012).. Do concurrent elections matter for explaining the level of invalid vote? Recent studies suggest that the answer is “yes.” Lysek et al. (2019) find that concurrent elections led to a substantial impact on invalid voting in post-communist democracies. Using data of presidential elections in post-communist and Latin American democracies between 1980 and 2013, Kouba. 1. Countries in Latin American that are regulated by the compulsory voting system are Argentina, Bolivia, Brazil, Costa Rica, Dominica Republic, Ecuador, Honduras post-1991, Mexico, Panama, Peru, and Uruguay. And those adopted a voluntary vote system are: Chile post-2012, Colombia, El Salvador, Guatemala post-1990, Nicaragua and, Venezuela post-1993.. 19 DOI:10.6814/NCCU202100292.

(20) and Lysek (2016) find that when lower chamber and presidential elections were held on the same date, invalid votes tend to be higher.. 2.2 Socioeconomic characteristics The second approach for explaining invalid voting focuses on the social structure and the economic capacity of voters to complete the vote. For instance, the level of invalid vote tends to be lower in highly urbanized areas because political information is easier to obtain (Power and Garand 2007). Pilet et al. (2019) found that invalid votes significantly increase in municipalities with a lower socio-economic performance. However, other studies find that urbanization is negatively associated with invalid votes (Ugla 2008; Zulfikarpasic 2001). Martins (2017) finds that. 政治大 out that unemployment rates have no significant effects on the invalid ballot. 立. unemployment in urban areas increases the rate of null and blank voting, while Ugla (2008) points. 學. ‧ 國. Economic performance can have significant effects on the political participation of citizens (Powell 1982) since it influences the nature of socio-economic factors such as level of. ‧. education, access to information, and levels of wealth which, in one way or another, transforms. sit. y. Nat. the relationships among different groups in society to create a diversity of interests.. io. er. All of these factors are most likely to increase the political participation of the citizens, affect voting behavior and political preferences. Relatedly, Power and Garand (2007) argued that higher. n. al. iv. per capita GDP increases personal economic n as improve trust in the political C satisfaction as well. hen. hi U. system, thus reducing the probability of castingg ancinvalid vote. However, this assumption for Cohen (2018, 406) has no statistical significance on the variation of the invalid vote since "policymotivated invalid voting is not driven, on average, by poor perceived economic outcomes, but rather by the perception that government performance has been poor across a range of policy areas."2 The level of education also matters for explaining electoral outcomes. For instance, existing studies have shown that education is strongly correlated to voter turnout (Driscoll and 2. The perception of good economic performance is scarce in Latin America (Corporación Latinobarómetro 2018). For 2018, according to Corporación Latinobarómetro (2018), around 35 percent of Latin Americans stated that the most critical economic issues within their respective countries are comprised of unemployment, low wages, and poverty.. 20 DOI:10.6814/NCCU202100292.

(21) Nelson 2014; Lysek et al. 2019; Power and Garand 2007; Tenn 2007). Educated citizens are more informed of the issues that are at stake in an election contest. They can read and comprehend voting and registration procedures (Driscoll and Nelson 2014), as well as likely having the awareness of considering voting as a civic duty (Tenn 2007). Martinez I Coma and Werner (2018) argued that voters with higher education levels seem to have a slight tendency to increase levels of invalid voting, as a way of intentionally expressing their discontent with the current political system (Zulfikarpasic 2001). However, other comparative studies reported a negative relationship between educational level and invalid vote (Aldashev and Mastrobuoni 2019; Blais 2006; Driscoll and Nelson 2014; Kouba and Lysek 2016; Power and Garand 2007). Less-educated voters3 tend to cast invalid votes because it would be. 政治大. more difficult for them to understand the politics and voting laws in particular. Similarly, Hill and. 立. Young (2007) and McAllister and Makkai (1993) showed that in Australia, the limited proficiency. ‧ 國. 學. of a country’s language (English) in immigrant communities make these “new” voters more likely. ‧ y. Nat. 2.3 Political context. sit. to spoil their votes.. io. er. The political context approach finds its argument in the form of protest against the performance of political institutions, due to dissatisfaction against the regime and political. n. al. iv. system (Cohen 2018; Driscoll and Nelson n and Werner 2018; Lysek et al. C 2014; Martinez I Coma. hen. hi U. g c and Garand (2007) found strong evidence 2019; Power and Garand 2007; Uggla 2008). Power that a negative change in levels of democracy is associated with a higher level of the invalid vote. At the individual level, Driscoll and Nelson (2014) report that spoiled ballots were driven by political concern where dissatisfaction with government was rampant. Cohen (2018) finds, using individual-level data across the Latin American region between 2008 and 2014, that perceptions of poor government performance and those who are more interested in politics on average might lead to more invalid voting. Similarly, Arbache et al. (2015) argue a low evaluation of political institutions is a powerful predictor of invalid voting. Power and Roberts’ (1995) study of twelve 3. In Latin America, a portion of voters is illiterate or semi-illiterate. For 2018, in this region, nearly 6.2 percent could not read or write (Statista 2020). Therefore, voting rights are restricted to these marginalized societal groups.. 21 DOI:10.6814/NCCU202100292.

(22) legislative elections in Brazil between 1945 and 1990 suggests that invalid voting rose when the regime was most discredited. One important political variable that might affect electoral outcomes is the issue of corruption. A classic and most widely used concept of corruption is given by The World Bank (1997) as the “misuse or the abuse of public office for private gain.” The Organization of Economic Cooperation and Development- OECD (2013) points out that corruption indicates an “active or passive misuse of the powers of Public officials (appointed or elected) for private financial or other benefits,” i.e., an illegal act committed by a public official who took advantage of the position and misused its power entrusted by law for personal gain and to provide favors to the. 政治大 Recent literature has suggested 立 that the level of corruption is associated with invalid. third person.. ‧ 國. 學. votes (e.g., Martinez I Coma and Werner 2018), further suggesting that a discontented individual is likely to cast a spoiled ballot as a rejection of a corrupted government. One main reason is that. ‧. the abuse of public office for private gain erodes people’s trust in government and institutions. In this sense, voter discontent and outright indignation have delivered in diverse expressions of. y. Nat. er. io. sit. dissatisfaction, and one type of these expressions is spoiling their votes. In Latin America, the growing a distrust and widespread discontent towards politicians by. n. iv l C n citizens have become a trend that has grown the exposure of corruption over the last h e nupgwith chi U. years (Galindo 2019). As it is presented in the Global Corruption Barometer by Transparency International (2019), around 57 percent of Latin American citizens think their government is not doing enough to end corruption. This argument is bolstered with the significant number of presidents and former presidents in the region who have been forced from office, jailed, or investigation of corruption cases. To address the research question of what explains the variation in invalid votes in Latin America, I argue that compulsory voting system, concurrent elections, level of economic performance, level of educational attainment, level of democracy, and corruption are possible factors in explaining the variation of invalid voting for presidential (first round) and lower. 22 DOI:10.6814/NCCU202100292.

(23) chamber elections in Latin America countries. To test the effects of each variable on the level of invalid vote, I generate the following hypotheses:. H1: A country tends to have more invalid votes in countries that adopt a compulsory voting system. H2: A country tends to have more invalid votes in elections when the presidential election and lower chamber elections are held on the same date (concurrent elections). H3: A country tends to have fewer invalid votes in countries with a better economic performance. H4: A country tends to have fewer invalid votes in countries where the citizens have a. 政治大 H5: A country’s invalid votes 立in elections decrease with a higher level of democracy.. higher level of educational attainment.. ‧. ‧ 國. 學. H6: A country’s invalid votes in elections increase with a higher level of corruption.. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. 23 DOI:10.6814/NCCU202100292.

(24) Chapter Three: Research Design. 3.1 Large- N Tests 3.1.1 Dependent Variable For this study, the unit of analysis is country-election. The dependent variable for this study is the percentage of invalid vote in an election. I gathered data from Nohlen (2005a; 2005b) and International IDEA (2019) for presidential elections and lower chamber elections in 18 Latin American countries from 1916 to 2018. For countries that adopt a presidential run-off electoral system, I used the results for the first-round election. Due to the unavailability of historical data,. 政治大 the observations included for the statistical analyses. 立. presidential and lower chamber elections for all countries start at different times. Table 1 shows. n. Bolivia. Brazil. y. sit. io. Lower Chamber Elections 1916, 1918, 1920, 1922, 1924, 1926, 1928, 1930, 1931, 1934, 1936, 1938, 1940, 1942, 1946, 1951, 1954, 1958, 1960, 1962, al v1963, 1965, 1973, 1983, i n Ch e n g c h i U 1985, 1987, 1989, 1991, 1993, 1995, 1997, 1999, 2001, 2003, 2005, 2007, 2009, 2011, 2013, 2015, 2017 1951,1956, 1960, 1964, 1966, 1956, 1958, 1960, 1962, 1978, 1979,1980, 1985, 1989, 1964, 1966, 1978, 1979, 1993, 1997, 2002, 2005, 1980, 1985, 1989, 1993, 2009, 2014 1997, 2002, 2005, 2009, 2014 1945, 1950, 1955, 1960, 1950, 1954, 1958, 1962, 1989, 1994, 1998, 2002, 1966, 1970, 1974, 1978, 2006, 2010, 2014, 2018 1982, 1986, 1990, 1994, 1998, 2002, 2006, 2010, 2014, 2018. er. ‧ 國. Nat. Presidential Elections 1916, 1922, 1928, 1931, 1937, 1946, 1951, 1958, 1963, 1973, 1983, 1989, 1995, 1999, 2003, 2007, 2011, 2015. ‧. Country Name Argentina. 學. Table 1: Observations Included for the Statistical Analyses. 24 DOI:10.6814/NCCU202100292.

(25) Chile. 1938, 1942, 1946, 1952,1958, 1964,1970, 1989,1993, 1999, 2005, 2009, 2013, 2017 1946, 1949, 1958, 1962, 1966, 1970, 1974, 1978, 1982, 1986, 1990, 1994, 1998, 2002, 2006, 2010, 2014, 2018. Colombia. Costa Rica. Dominican Republic. ‧ 國. n. er. io. sit. y. ‧. Nat. Guatemala. 政治大. 學. El Salvador. 1951, 1953, 1958, 1960, 1962, 1964, 1966, 1968, 1970, 1974, 1978, 1982, 1986, 1990, 1991, 1994, 1998, 2002, 2006, 2010, 2014, 2018 1953, 1958, 1962, 1966, 1953, 1958, 1962, 1966, 1970, 1974, 1978, 1982, 1970, 1974, 1978, 1982, 1986, 1990, 1994, 1998, 1986, 1990, 1994, 1998, 2002, 2006, 2010, 2014, 2018 2002, 2006, 2010, 2014, 2018 1962, 1966, 1970, 1974, 1962, 1966, 1970, 1974, 1978, 1982, 1986, 1990, 1978, 1982, 1986, 1990, 1994, 1996, 2000, 2004, 1994, 1998, 2002, 2006, 2008, 2012, 2016 2010, 2016 1968, 1978, 1984, 1988, 1962, 1979, 1984, 1986, 1992, 1996, 1998, 2002, 1988, 1990, 1992, 1994, 2006, 2009, 2013, 2017 1996, 1998, 2006, 2009, 2013, 2017 1984, 1989, 1994, 1999, 1985, 1988, 1991, 1994, 2004, 2009, 2014 1997, 2000, 2003, 2006, 2009, 2012, 2015, 2018 1958, 1966, 1970, 1974, 1959, 1961, 1966, 1970, 1982, 1985, 1990, 1995, 1985, 1990, 1994, 1999, 1999, 2003, 2007, 2011, 2015 2003, 2007, 2011, 2015 a l 1954, 1971, 1981, i v1981, 1985, 1989, 1993, 1948, n C h e1993, 1985, 1989, 1997, i U 1997, 2001, 2005, 2009 h n c g 2001, 2005, 2009, 2013, 2017 2013, 2017, 1982, 1988, 1994, 2000, 1970, 1973, 1976, 1979, 2006, 2012, 2018 1982, 1985, 1988, 1991, 1994, 1997, 2000, 2003, 2006, 2009, 2012, 2015, 2018 1984, 1990, 1996, 2001, 1990, 1996, 2001, 2006 2006, 2011, 2016 1948, 1952, 1956, 1964, 1978, 1984, 1994, 1999, 1968, 1984, 1989, 1994, 2004, 2009, 2014 1999, 2004, 2009, 2014 1958, 1963, 1968, 1973, 1963, 1968, 1973, 1978, 1978, 1983, 1988, 1989, 1983, 1988, 1989, 1993, 1993, 1998, 2000, 2003, 1998, 2003, 2008, 2013, 2018 2008, 2013, 2018. 立. Ecuador. Honduras. Mexico. Nicaragua Panama. Paraguay. 1989, 1993, 1997, 2001, 2005, 2009, 2013, 2017. 25 DOI:10.6814/NCCU202100292.

(26) Peru. 1956, 1962, 1963, 1980, 1985, 1990, 1995, 2000, 2001, 2006, 2011, 2016 1984, 1989, 1994, 1999, 2004, 2009, 2014 1958, 1963, 1968, 1973, 1978, 1983, 1988, 1993, 1998, 2000, 2006. Uruguay Venezuela. 1980, 1985, 1990, 1995, 2000, 2001, 2006, 2011, 2016 1971, 1984, 1989, 1994, 1999, 2004, 2009, 2014 1947, 1958, 1963, 1968, 1973, 1978, 1983, 1988, 1993, 1998, 2000, 2005. Note: Due to a lack of availability of reliable data, I do not include some electoral years from Venezuela and Nicaragua in this research.. 3.1.2 Independent Variables. 治政 approaches for invalid votes: institutional framework, 大 socioeconomic characteristics, and 立 political context. For the Institutional variables, I include two variables about compulsory voting As discussed before, this thesis will test variables from three main explanatory. ‧ 國. 學. system. It will be operationalized as three categorical variables: voluntary voting system, unenforced mandatory voting system (without fines or legal sanctions), and enforced compulsory. ‧. voting (with fines and legal sanctions). The data are from the Compulsory voting index from V-. Nat. sit. y. Dem Dataset - Version 9 (Coppedge et al. 2019; Pemstein et al. 2019). In my empirical models, I. er. io. include unenforced mandatory voting system and enforced compulsory voting system and leave. n. the voluntary voting system as theareference group. In addition,vI also include a variable for the. l. i. n as "1" when presidential and concurrent elections. This variable will C beh a dummy variable Ucoded engchi. lower chamber election were held on the same date, and 0 otherwise. The data are from Nohlen (2005a; 2005b) and International IDEA (2019). For the socioeconomic variables, I include the level of economic development, which is operationalized as GDP per capita (log). I also include GDP growth as a variable for economic performance. The data are from V-Dem Dataset - Version 8 (Coppedge et al. 2018; Pemstein et al. 2018). Moreover, I include the level of education in the models. I use the Education15+ Index from the V-Dem Dataset - Version 9 (Coppedge et al. 2019; Pemstein et al. 2019), which is operationalized as taking the average years of educational attainment among citizens who are older than 15. 26 DOI:10.6814/NCCU202100292.

(27) For the variables regarding political context, I first include the level of democracy, using the Polity revised combined score from the V-Dem Dataset - Version 9 (Coppedge et al. 2019; Pemstein et al. 2019). This variable is a modified version of the Polity variable added to facilitate the use of the Polity regime measure in time-series analyses. Last, I test the effects of corruption on invalid vote. I use the Regime Corruption index by the V-Dem Dataset - Version 9 (Coppedge et al. 2019; Pemstein et al. 2019). The main question of this index is: To what extent do political actors use political office for private or political gain? The index is formed by taking the reversed point estimates (so that higher scores = more regime corruption) from a Bayesian factor analysis model of the indicators for executive embezzlement, executive bribes, legislative corruption, and judicial corruption.. 3.1.3 Control Variables. 立. 政治大. ‧ 國. 學. In addition to the independent variables that I mentioned above, I control for a dummy variable for founding election. According to Bogdanor (1990), founding elections are often known. ‧. as the endpoint of democratic transition when democratic governments replace authoritarian regimes. For Power and Garand (2007), it is expected that particularly in founding elections, the. y. Nat. sit. variation of invalid ballots to be lower than in other polls because of those elections are the first. er. io. free and democratic contest after the end of authoritarian power. The founding election variable. n. a is the first election in a iregime is coded as "1" when the election period after the end of v. l C n U hen authoritarian rule, and 0 otherwise. The data for thish variable are from Nohlen (2005a; 2005b), i gc Power et al. (2004), and Lucardi (2006). Table 2 provides a short description for the variables used in the statistical analyses. And, Tables 3 and 4 provide summary statistics of the variables used in the empirical analysis. .. 27 DOI:10.6814/NCCU202100292.

(28) Table 2: List of Variables. 學. Measurements Percentage of invalid voting in the first round of Presidential election and the unicameral national parliaments or lower chamber of the national legislatures. Enforced compulsory voting A dummy variable coded as “1” 治政 system for countries 大 with an enforced 立 compulsory voting system, and “0” otherwise. Unenforced compulsory voting A dummy variable coded as “1” system for countries with an unenforced compulsory voting system, and “0” otherwise. Concurrent elections A dummy variable coded as “1” when presidential and Lower al iv chamber nelection were held on Ch U i e n gthe c h same date, and “0” otherwise. GDP per capita Natural log of GDP per capita. ‧. io. sit. y. Nat. n. er. Independent Variables. Labels Invalid vote. ‧ 國. Variables Dependent Variable. GDP growth. Percentage of GDP growth. Sources Nohlen (2005a; 2005b) and International IDEA (2019).. The Compulsory voting index from the V-Dem Dataset Version 9 (Coppedge et al. 2019; Pemstein et al. 2019). The Compulsory voting index from the V-Dem Dataset Version 9 (Coppedge et al. 2019; Pemstein et al. 2019). Nohlen (2005a; 2005b) and International IDEA (2019).. V-Dem Dataset – Version 8 (Coppedge et al. 2018; Pemstein et al. 2018) V-Dem Dataset – Version 8 (Coppedge et al. 2018; Pemstein et al. 2018). 28 DOI:10.6814/NCCU202100292.

(29) Level of education. n. er. io. sit. y. Nat. Control Variables. ‧. ‧ 國. 學. Percentage of average years of The data is from the education among citizens older Education15+ Index from the than 15 V-Dem Dataset - Version 9 (Coppedge et al. 2019; Pemstein et al. 2019) Level of democracy The “Polity Score” evaluates The Polity revised combined the type of regime in a country score from the V-Dem Dataset with a range from -10 - Version 9 (Coppedge et al. (hereditary monarchy) to +10 2019; Pemstein et al. 2019) (consolidated democracy). This variable 治 is lagged by one year. 政大 Level of corruption The “Regime Corruption” Regime Corruption index from 立 examines to what extent do the V-Dem Dataset - Version 9 political actors use political (Coppedge et al. 2019; office for private or political Pemstein et al. 2019) gain. An interval from low to high (0-1) Founding election A dummy variable coded as “1” Nohlen (2005a; 2005b), Power when it is the first election in a et al. (2004), and Lucardi regime period al v after the end of (2006). i n authoritarian rule, and “0” Ch e n gotherwise. chi U. 29 DOI:10.6814/NCCU202100292.

(30) Table 3. Summary Statistics of the Variables Used in the Empirical Analyses for presidential election Variables Invalid vote. St. Dev 5.49813 0.5007958. Min 0.01 0. Max 38.37 1. Obs 226 226. 0.2831858. 0.451546. 0. 1. 226. 0.7743363. 0.4189466. 0. 1. 226. 8.567104. 0.6505808. 7.102. 9.973. 221. 0.0243439. 0.0552165. -0.183. 0.282. 221. 5.906093. 1.947284. 1.64. 10.36. 226. 10. 226. 4.380531. ‧ 國. 立 0.5895752 0.1327434. 治政 5.641621 -9 大 0.260299. 0.02. 0.963. 226. 0.3400503. 0. 1. 226. 學. Unenforced compulsory voting system (H1) Enforced compulsory voting system (H1) Concurrent Elections (H2) Logged per capita GDP (H3) GDP growth (H3) Educational attainment (H4) Level of democracy (H5) Regime corruption (H6) Founding elections. Mean 5.857655 0.4823009. Obs 255 255. 1. 255. y. Max 46.12 1. sit. Min 0.05 0. n. al. St. Dev 8.266502 0.500887. er. Mean 8.649529 0.5098039. io. Unenforced compulsory voting system (H1) Enforced compulsory voting system (H1) Concurrent Elections (H2) Logged per capita GDP (H3) GDP growth (H3) Educational attainment (H4) Level of democracy (H5) Regime corruption (H6) Founding elections. Nat. Variables Invalid vote. ‧. Table 4. Summary Statistics of the Variables Used in the Empirical Analyses for lower chamber election. C h0.4451374 0 U n i 0.2705882 engchi. v. 0.6196078. 0.486438. 0. 1. 255. 8.680743. 0.6416347. 7.127. 9.973. 249. 0.0256948. 0.0542779. -0.183. 0.282. 249. 6.010302. 1.915281. 1.808. 10.36. 255. 4.364706. 5.728463. -9. 10. 255. 0.5839294. 0.2495473. 0.02. 0.963. 255. 0.1098039. 0.31326. 0. 1. 255. 30 DOI:10.6814/NCCU202100292.

(31) 3.2 Methods and Estimation Techniques In this research, I employ a mixed-method (quantitative and qualitative methods) approach for empirical analyses. The mixed-method seeks data from as many different sources as possible and employs more than one way to secure and validate the data. Besides, it provides more rigorous results and a better understanding of the research problem than any approach alone (Haq 2014). For the quantitative analyses, I employ two estimation techniques. First, I use ordinary least squares (OLS) regressions to estimate invalid vote. Second, I use pooled time-series cross-sectional (TSCS) analyses with panel-corrected standard errors (Beck and Katz 1995). To further explore the mechanisms behind the quantitative findings (Grinnell and Unrau. 政治大 the qualitative analyses. Why do campaigns for invalid voting matter? Jacobson (2015) argued 立 2010), I conduct comparative case studies of campaigns that encourage casting invalid votes for. ‧ 國. 學. that information provided by campaigns affect the vote decision in different ways. In general, the electoral campaigns persuade people to vote for a particular candidate, but the campaigns can. ‧. also be mobilized for advocating not to cast an invalid ballot. Relatedly, campaigns in favor of invalid votes are directed not only to voters with party identification but mainly to voters who. y. Nat. sit. want to exercise their right to abstain (Cisneros 2012). Crespo (2010) explains that these social. n. al. er. io. movements push the idea of a total boycott by using a null vote rather than abstaining.. iv. For the comparative case studies, n campaigns during presidential C I examine invalid voting. hen. hi U. c elections and lower chamber elections in LatingAmerica based on Internet sources, journals, books, newspapers, and other mass media. I analyze the cross-national similarities and differences in the strategic settings of the campaigns such as election round, compulsory voting system, concurrent elections, justification of campaigns, leadership, and results. Although there are previous country-level studies on invalid voting campaigns in Latin America (e.g., Cisneros 2012; Cisneros and Freigedo 2014; D' Adamo and Garcia 2003; Sonnleitner 2019), regional comparative evidence is scarce (Cohen 2016). Therefore, the qualitative analyses of this research aims to fill the gap in the literature for providing a better understanding of the behavioral dynamics of the invalid vote.. 31 DOI:10.6814/NCCU202100292.

(32) Chapter Four: Quantitative Analysis. 4.1 Invalid Voting in Latin America Over the past decades, mass popular uprisings, the resignation of presidents, populist governments, failed attempts at coups d'état, fraudulent elections, high rates of corruption, among others, have drawn the political panorama in Latin America. Despite the role played by "the third wave" in re-establishing democracy with a well-institutionalized representative election, many citizens have preferred a "candidate without a name," a null or blank option as a representation of their preferences.. 立. 政治大. Invalid vote has become the centerpiece of these citizens’ protest. As shown in Figure 5. ‧ 國. 學. and Figure 6, we see an alarmingly increasing in invalid vote in some of the Latin American electoral processes either in the presidential election (like Brazil, Ecuador, Guatemala, Honduras,. ‧. Peru, and Paraguay) or in Lower chamber election (like Bolivia, Brazil, Chile, Colombia, Ecuador,. n. al. er. io. sit. y. Nat. Guatemala, Nicaragua, Panama, Paraguay, and Peru).. Ch. engchi. i n U. v. 32 DOI:10.6814/NCCU202100292.

(33) Figure 5: Percentage of Invalid Vote in presidential elections in Latin America. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. n engchi U. iv. 33 DOI:10.6814/NCCU202100292.

(34) 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. n engchi U. iv. Source: Author. Data are from Nohlen (2005a; 2005b) and International IDEA (2019).. 34 DOI:10.6814/NCCU202100292.

(35) Figure 6: Percentage of Invalid Vote in Lower Chamber elections in Latin America. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. n engchi U. iv. 35 DOI:10.6814/NCCU202100292.

(36) 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. n engchi U. iv. Source: Author. Data are from Nohlen (2005a; 2005b) and International IDEA (2019).. 36 DOI:10.6814/NCCU202100292.

(37) Relatedly, since the re-democratization of Latin America, countries had survived at all under the stress of several reforms from authoritarian practices and political scandals. Many of these new regimes showed great optimism in the implementation of new measures of legitimizing power while ensuring a wider range of democratic functions. Increasing citizen's competence and participation in the electoral process were some of them. Precisely, the reality is that in the region today, several factors seem to affect the proportion of voters that cast invalid ballots. In this vein, I test hypotheses regarding compulsory voting systems, concurrent elections, economic development and performance, educational attainment, level of democracy, and corruption on invalid voting. Existing works on the invalid. 政治大. voting in Latin America have focused on some of these factors (e.g. Cohen 2018; Martinez I Coma and Werner 2018; Power and Garand 2007; Uggla 2008); however, not all of them cover a large. 立. dataset combining the qualities of cross-sectional with time-series data from presidential (first. ‧ 國. 學. round) and lower chamber elections and invalid voting campaigns, which is what I will do in the. ‧ y. Nat. 4.2 Empirical Results. sit. next section.. io. er. Tables 5 shows the empirical results for the tests of my six hypotheses for Latin American presidential elections from 1916 to 2018. As can be seen, the results in Model 1 (OLS) and Model. n. al. iv. 2 (PCSE) are largely similar. RegardingC the tests of the institutional hypotheses, I find that the n. hen. hi U. g cis positive and statistically significant. This coefficient of enforced compulsory voting system finding suggests that invalid voting tends to be higher in countries with enforced compulsory. voting system, which is similar to the results of previous studies (Martinez I Coma and Werner 2018; Power and Garand 2007). Substantively, this finding suggests that compared to a country with voluntary voting system, the invalid voting rates in the presidential elections is 3.9% higher for a country with enforced compulsory system.. 37 DOI:10.6814/NCCU202100292.

(38) Table 5. Results for Model of Invalid Voting in Latin America Presidential elections Model 1 Model 2 Explanatory Variables (Ordinary Least Squares (Panel-Corrected Standard (OLS)) Error (PCSE)) Unenforced compulsory -1.734 -1.734* voting system (H1) (1.127) (0.971) Enforced compulsory voting 3.893*** 3.893*** system (H1) (1.282) (1.143) Concurrent Elections (H2) 2.413*** 2.413** (0.888) (1.141) Logged per capita GDP (H3) 0.361 0.362 (0.759) (1.533) GDP growth (H3) -9.618* -9.618 (5.753) (7.686) Educational attainment (H4) -0.265 -0.265 (0.229) (0.425) Level of democracy (H5) -0.016 -0.016 (0.869) (0.070) Regime corruption (H6) 3.422*** 3.422** (1.141) (1.607) Founding elections -0.546 -0.545 (1.218) (1.602) a v i Constant 0.565 l C 0.5653 n h e(6.486) (11.695) ngchi U Observations (N) 221 221 R-Squared 0.214 0.214 Notes: Voluntary voting system is the reference category for voting systems. Robust standard errors are in parentheses. *p<0.1; **p<0.05; ***p<0.01.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. In addition, the coefficient for the concurrent elections is also positive and statistically significant. Substantively, the finding suggests that compared to a country that holds presidential election and lower chamber election on different dates, the invalid voting rates in the presidential elections is 2.4% higher for a country that holds presidential election and lower chamber election. 38 DOI:10.6814/NCCU202100292.

(39) on the same date. This result is similar to Kouba and Lysek’s (2016) analysis. Third, my results show that the coefficient for the regime corruption is positive and statistically significant, suggesting that voters in a country with a higher level of regime corruption tends to cast more invalid votes in the presidential elections. Theoretically, this finding might indicate that voters tend to cast invalid votes as a mean of protest against the corrupt government. In short, my analyses provide strong support for H1, H2, and H6. As for other explanatory variables, Table 5 shows mixed results for unenforced compulsory voting system and GDP growth. The coefficient of unenforced compulsory voting system is negative in Model 1 and Model 2, but it only reaches statistical significance for the. 政治大. Model 2 (p=0.074) but not Model 1. A t-test shows that the coefficients of this variable in both models are not significantly different from one another. Moreover, the coefficient of GDP growth. 立. is negative in both Model 1 and Model 2, suggesting that a country might have a lower level of. ‧ 國. 學. invalid votes if the economic performance is better. However, the coefficient of GDP growth reaches statistical significance for the Model 1 (p=0.096) but not Model 2. A t-test shows that the. ‧. coefficients of GDP growth in both models are not significantly different from one another. This. sit. y. Nat. result suggests that in Latin America, poor economic performances might not drive popular discontent to cast invalid votes for presidential elections.. er. io. n. a l variables do not have statistically Surprisingly, other explanatory significant effects on iv Un. C. h eelection. i invalid votes in my sample of presidential to theoretical expectation, the n g c hContrary coefficient of per capita GDP is positive, suggesting that a country with a higher level of economic development tends to have a higher level of invalid voting rates. However, the coefficient of this variable is not statistically significant. On this point, a probable explanation can be discussed. As Martins (2017) says that, people with economic hardships, who still want to vote, are more likely to spoil their vote as a way to manifest their discontent regarding the government’s policies; however, this is not unrelated to the fact that there may be little participation of the poor and the unemployed in the elections, who often exhibit a strong decline in their ability to take part in politics and, of course, it is not reflected in the percentage of invalid votes.. 39 DOI:10.6814/NCCU202100292.

(40) Educational attainment has a coefficient that fails to reach conventional levels of statistical significance. Contrary to Power and Garand (2007), that rejects the idea that in Latin America, a higher level of education might turn out a lower rate of invalid votes. One possible answer to explain this lack of correlation between education and invalid votes is given in that educational levels might be related to two different types of invalid voters, one with a strong political dissatisfaction and intentionally spoil the ballots and the other who does not have enough voting skills or is misinformed about politics and unintentionally spoil the vote. Moreover, for Models 1 and 2, the level of democracy is negatively correlated with invalid votes. However, the effect of democracy is not statistically significant, suggesting that the effect is no different from zero. This finding shows that the relationship between the level of democracy and. 政治大 high or low in founding election. In 立short, the results provide no statistically significant support. invalid voting might be rather complicated. Last, I find that invalid voting rates are not particularly. ‧ 國. 學. for H3, H4, and H5.. In Table 6, I demonstrate the statistical results about lower chamber elections. The results. ‧. in Model 3 and Model 4 are similar. Consistent with the findings for the presidential elections, I. sit. y. Nat. find that countries that adopt enforced compulsory voting system tend to have higher levels of. io. er. invalid vote than countries that adopt voluntary voting system. Moreover, the results indicate that invalid voting rates tend to be higher in countries that suffer from higher levels of regime. n. al. corruption.. Ch. engchi. i n U. v. Table 6. Results for Model of Invalid Voting in Latin America lower chamber elections. Explanatory Variables Unenforced compulsory voting system (H1) Enforced compulsory voting system (H1) Concurrent Elections (H2). Model 3 (Ordinary Least Squares (OLS)) -3.357*** (1.242) 8.107*** (1.665) -0.724 (1.050). Model 4 (Panel-Corrected Standard Error (PCSE)) -3.357*** (1.051) 8.107*** (1.448) -0.724 (0.967). 40 DOI:10.6814/NCCU202100292.

(41) Logged per capita GDP (H3) GDP growth (H3) Educational attainment (H4) Level of democracy (H5) Regime corruption (H6) Founding elections Constant. 立. 3.262*** (1.232) -0.908 (9.521) -0.057 (0.360) 0.055 (0.088) 7.043*** (1.838) -0.328 (1.490) -24.099** (9.646) 249 0.308. 3.262*** (1.238) -0.908 (7.818) -0.572 (0.291) 0.055 (0.078) 7.043*** (1.932) -0.328 (1.408) -24.099** (10.183) 249 0.308. 政治大. ‧. ‧ 國. 學. Observations (N) R-Squared Notes: standard errors are in parentheses. *p<0.1; **p<0.05; ***p<0.01.. Interestingly, for lower chamber elections, the results show that the coefficient for. sit. y. Nat. unenforced compulsory voting system is negative and statistically significant. This finding suggests that countries that adopt unforced compulsory voting systems are more likely to have. er. io. n. lower levels of invalid voting rates than a countries that adopt voluntary voting system. In addition,. l. iv. n C h and statisticallyUsignificant. the coefficient for per capita GDP is positive This finding suggests that engchi. countries with higher levels of economic development tend to have higher levels of invalid vote. In short, the analyses of lower chamber elections provide strong support for H1, H3, and H6. Regarding other explanatory variables, I find that the coefficient for concurrent elections is negative and statistically insignificant. Similarly, the coefficients for GDP growth and educational attainment are also negative and statistically insignificant. The coefficient for the level of democracy is positive, but it does not reach statistical significance. Last, consistent with what I have found for the presidential elections, founding election does not have a strong impact on invalid votes. In other words, there is not sufficient evidence that first free and democratic. 41 DOI:10.6814/NCCU202100292.