國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
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 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 dataset combining the qualities of cross-sectional with time-series data from presidential (first round) and lower chamber elections and invalid voting campaigns, which is what I will do in the next section.
4.2 Empirical Results
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 2 (PCSE) are largely similar. Regarding the tests of the institutional hypotheses, I find that the coefficient of enforced compulsory voting system is positive and statistically significant. This 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.
‧
Table 5. Results for Model of Invalid Voting in Latin America Presidential elections
Explanatory Variables Concurrent Elections (H2) 2.413***
(0.888)
2.413**
(1.141) Logged per capita GDP (H3) 0.361
(0.759) Educational attainment (H4) -0.265
(0.229)
-0.265 (0.425)
Level of democracy (H5) -0.016
(0.869)
-0.016 (0.070)
Regime corruption (H6) 3.422***
(1.141)
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.
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
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
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 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 is negative in both Model 1 and Model 2, suggesting that a country might have a lower level of 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 result suggests that in Latin America, poor economic performances might not drive popular discontent to cast invalid votes for presidential elections.
Surprisingly, other explanatory variables do not have statistically significant effects on invalid votes in my sample of presidential election. Contrary to theoretical expectation, the 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.
‧
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 invalid voting might be rather complicated. Last, I find that invalid voting rates are not particularly high or low in founding election. In short, the results provide no statistically significant support 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 find that countries that adopt enforced compulsory voting system tend to have higher levels of 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 corruption.
Table 6. Results for Model of Invalid Voting in Latin America lower chamber elections
Explanatory Variables Concurrent Elections (H2) -0.724
(1.050)
-0.724 (0.967)
‧
Logged per capita GDP (H3) 3.262***
(1.232) Educational attainment (H4) -0.057
(0.360)
-0.572 (0.291)
Level of democracy (H5) 0.055
(0.088)
0.055 (0.078)
Regime corruption (H6) 7.043***
(1.838)
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 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 lower levels of invalid voting rates than countries that adopt voluntary voting system. In addition, the coefficient for per capita GDP is positive and statistically significant. This finding suggests that 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
‧
elections after the end of authoritarian power tended to have lower invalid vote rates. In short, the empirical results provide no statistically significant support for H2, H4, and H5.
To better understand whether different explanatory variables have different impacts on invalid voting in presidential elections and lower chamber elections, I conduct t-tests for comparing coefficients for variables that have statistically significant coefficient in either Model 1 or Model 3, or both.
Table 7. Consistency of Results for Presidential Elections and Lower Chamber Elections
Explanatory Variables
Insignificant Significant 0.968 Enforced compulsory voting system
(H1)
Significant Significant 2.005 Concurrent Elections (H2) Significant Insignificant 3.483 Logged per capita GDP (H3) Insignificant Significant -2.005
Regime corruption (H6) Significant Significant -1.674
Note: The T-test statistic must have an absolute value of at least 1.96 to be significant at p<0.05 and of at least 1.65 to be at p<0.10.
As can be seen in Table 7, the difference in the coefficients for three explanatory variables is statistically significant for presidential elections and lower chamber elections. First, while a country with enforced compulsory voting system tends to have higher levels of invalid vote for both presidential elections and lower chamber elections, the T-test shows that the difference in the effects of this type of voting system is statistically significant. In other words, a country with an enforced compulsory voting system tends to have higher levels of invalid vote for lower chamber elections than presidential elections. In this scenario, when voters are mandated to go to the polls while facing heavy penalties imposed if they do not vote, the next step is to think for whom to vote. Presidential elections often focus more attention than legislative elections (Monroy 2012), perhaps due to the excessive presence of the presidential figure in the political history of Latin America and the general perception of Latinos that the country's politics depend
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
fundamentally on the President and not of deputies. Therefore, the idea of electing a presidential candidate is probably greater and the percentage of invalid votes is reduced, while it might be the opposite effect for lower chamber elections.
The T-test statistic for comparing coefficients in regime corruption exhibit similar results.
Specifically, the results show that while the level of regime corruption has a positive effect on increasing invalid votes, such effect is stronger for lower chamber elections than presidential elections. A possible explanation for this argument is due to the fact that the generalized perception of Latinos towards the corruption of diverse political and social actors, falls mainly on the figure of parliamentarians, as noted by Corporación Latinobarómetro (2018) in which "51%
of Latin Americans believe that "all or almost all" parliamentarians are involved in acts of corruption", while 50% fell on the presidential figure.
Table 7 also shows interesting T-test statistics for concurrent elections and level of economic development. I find that the differences in coefficients of concurrent elections and per capita GDP for presidential elections and lower chamber elections are statistically significant.
Taking into account the results in Model 1 and Model 3, the finding suggests that concurrent elections tend to increase invalid vote for presidential elections but not for lower chamber elections. Moreover, the finding suggests that a higher level of economic development tends to increase invalid vote for lower chamber elections but not for presidential elections. Last, although the results in Model 1 and Model 3 show that the coefficient of unenforced compulsory voting system is statistically significant for lower chamber elections but statistically insignificant for presidential elections, the T-test statistic shows that the difference is not statistically significant.