國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
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 more difficult for them to understand the politics and voting laws in particular. Similarly, Hill and Young (2007) and McAllister and Makkai (1993) showed that in Australia, the limited proficiency of a country’s language (English) in immigrant communities make these “new” voters more likely to spoil their votes.
2.3 Political context
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 system (Cohen 2018; Driscoll and Nelson 2014; Martinez I Coma and Werner 2018; Lysek et al.
2019; Power and Garand 2007; Uggla 2008). Power and Garand (2007) found strong evidence 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
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
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 third person.
Recent literature has suggested that the level of corruption is associated with invalid 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 dissatisfaction, and one type of these expressions is spoiling their votes.
In Latin America, the growing distrust and widespread discontent towards politicians by citizens have become a trend that has grown up with the exposure of corruption over the last 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
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
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 higher level of educational attainment.
H5: A country’s invalid votes in elections decrease with a higher level of democracy.
H6: A country’s invalid votes in elections increase with a higher level of corruption.
‧
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, presidential and lower chamber elections for all countries start at different times. Table 1 shows the observations included for the statistical analyses.
Table 1: Observations Included for the Statistical Analyses
Country Name Presidential Elections Lower Chamber Elections
Argentina 1916, 1922, 1928, 1931,
1937, 1946, 1951, 1958, 2009, 2011, 2013, 2015, 2017
Bolivia 1951,1956, 1960, 1964, 1966,
1978, 1979,1980, 1985, 1989, 1997, 2002, 2005, 2009, 2014
Brazil 1945, 1950, 1955, 1960,
1989, 1994, 1998, 2002,