Who Votes in Latin America? A Test of Three Theoretical Perspectives

Abstract

Previous studies of electoral participation in Latin America have focused on the political and institutional factors that influence country differences in the aggregate level of turnout. This article provides a theoretical and empirical examination of the individual-level factors that have an impact on citizens’ propensity to vote. We test three theoretical perspectives that have been used to explain electoral participation in industrialized democracies (voters’ resources, voters’ motivations, and mobilization networks). Using a series of logistic and hierarchical models, we demonstrate that the demographic characteristics of voters (age and education) and citizens’ insertion in mobilizing networks (civic organizations and the working place) are strong predictors of electoral participation in Latin America. Our analysis also confirms the importance of contextual and institutional variables to explain turnout in the region.

Keywords

electoral participation resources education social capital Latin America

Introduction

After 30 years of uninterrupted democratic rule in most Latin American countries, and despite the clear normative and political consequences of electoral participation (Lijphart, 1997; Pateman, 1976; Pitkin & Shumer, 1982), we still know very little about the factors that affect individuals’ propensity to vote in Latin America. Of course, we are not the first scholars interested in turnout in Latin American countries. Over the last decade, several scholars have investigated the determinants of the cross-country differences in electoral participation (Fornos, Power, & Garand, 2004; Pérez-Liñán, 2001). These studies have demonstrated that a series of institutional and contextual factors have a positive impact on turnout. Electoral participation increases when registration procedures are efficient, when voting is compulsory and sanctions for abstaining are enforced, and when legislative and presidential elections are held concurrently. Turnout also tends to be higher in transitional elections (Kostadinova & Power, 2007). Other scholars have analyzed the differences in electoral participation within countries, comparing turnout rates in different regions or municipalities (e.g., Lehoucq & Wall, 2004; Remmer, 2010). These studies are informative about the institutional and contextual factors that influence turnout in specific countries. But they are not designed to address the question that interests us here, namely, the individual resources and motivations that increase the probability of voting in Latin America. In fact, all these studies analyze aggregate levels of electoral participation—at the national, regional, or local level. The conventional wisdom holds that socioeconomic factors are not related with aggregate turnout in Latin American countries (Fornos et al., 2004). Similarly, the studies of turnout at the subnational level have found inconsistent evidence for the impact of variables such as literacy, wealth, and population age on electoral participation.

In this article, we reassess the link between sociodemographic characteristics and turnout at the individual level with recent survey data from 18 Latin American countries. We also evaluate how citizens’ motivations and insertion in mobilizing networks affect their likelihood of going to the polls. We find evidence that the individual incentives to vote in Latin America are driven mainly by individual sociodemographic attributes and insertion in mobilization networks. In particular, this article demonstrates that older, educated, politically motivated, and civic-active citizens are more likely to vote.

This article will proceed as follows. First, we present the theoretical expectations guiding this research. We introduce three explanations for voter turnout that have dominated the study of electoral participation in Western democracies: voters’ resources, voters’ motivations, and networks of mobilization. Second, we present the data and the model estimation. Third, we describe the statistical results. The following section goes further by discussing which variables are the strongest predictors of electoral participation in Latin America, and which one of the three theoretical perspectives is more relevant in the Latin American context. The final section concludes and suggests avenues for further research.

Theory

The literature on electoral participation in Europe and the United States is immense.¹ This literature has analyzed the impact of dozens of variables on the likelihood that an individual will turn out on election day. In a seminal book, Verba, Schlozman, and Brady (1995) created a useful theoretical framework to explain individual’s political participation. They started off by inverting the traditional question and asked why some people do not take part in politics. Their framework suggests three different answers to this question: They can’t because they lack necessary resources, they don’t want to because they lack sufficient interest or knowledge, and nobody asked them to because they fall outside of the traditional networks that rally voters. In other words, their understanding of the participatory process rests on three main factors: motivation, capacity, and networks of recruitment.

Verba et al. (1995) used this model to explain different forms of political participation in the United States. But this framework was explicitly designed as a “road map for the understanding of political participation in any democracy” (Verba et al., 1995, p. 25). Given the dearth of previous research on electoral participation in Latin America at the individual level, in this article, we make an important empirical contribution by assessing the validity of these different perspectives in the Latin American context. Unlike other scholars (Fornos et al., 2004), we do not expect Latin American citizens to behave in fundamentally different ways from American and European voters. In other words, we expect that a series of variables associated with these three perspectives (motivation, capacity, and networks of recruitment) will also be correlated with electoral participation in Latin America. But the gist of our empirical contribution attempts to assess which one of these broad categories of reasons explains more. In other words, this article is not so much about whether these perspectives explain electoral participation in Latin America but rather about how much they explain.

Resources and Voters’ Capacity

Verba et al. (1995) argue that voting is a unique form of political engagement that is less demanding in resources than working in campaigns, writing letters to government officials, or donating money to party activities. Although voting requires less time and money than other political activities, citizens still need a minimum of skills and resources to understand what is at stake and to gain interest in the outcome of the election.

The socioeconomic status (SES) model of voter turnout has consistently shown that income and education are positively associated with electoral participation at the individual level. Individuals with a higher SES are more likely to turn out than poorer and less educated citizens (Leighley & Nagler, 1992; Verba, Nie, & Kim, 1978; Wolfinger & Rosenstone, 1980). These studies suggest that citizens with higher SES tend to have more free time to participate in political activities and are better informed. More educated individuals are also more likely to understand the issues at stake in the elections and to become politically interested (Brady, Verba, & Schlozman, 1995). We expect then that the probability of voting increases when the level of education increases.

Controlling for education, the level of income should be less directly related to electoral participation, because voting requires minimal monetary resources. Still, going to the polling station may require that citizens take some form of public transportation. Even these minimal expenses may be prohibitive for the more destitute voters, especially if they are not registered to vote in the place where they live. Hence, we expect a difference in the likelihood of voting between the poorest voters and the rest of the population, but we do not expect a linear relationship between income and turnout.

Another essential individual resource is political experience. Many studies demonstrate that older citizens tend to vote more than their younger counterparts. Previous research has found strong support for this relationship at the individual level both in developed (Leighley & Nagler, 1992; Wolfinger & Rosenstone, 1980) and in developing countries (Niemi & Barkan, 1987). In fact, political socialization takes time. Young voters may be disoriented by the different electoral options, thereby preferring not to vote. Political experience is acquired over time as citizens face concrete policy issues (e.g., housing, taxation, health, social benefits), discuss about politics in the workplace or in their social networks, and learn about the different programs political parties propose to solve the problems they face. This process can take several years. Hence, we expect that more experienced Latin American voters (i.e., older citizens) tend to vote more than political neophytes (i.e., younger voters).

Motivation

Not all citizens who have the capacity to vote go to the polls on election day. A key explanatory factor in Verba et al.’s (1995) model is motivation. Individuals who are interested in political debates and who have enough political knowledge to understand what is at stake are more likely to participate in elections.

Survey evidence demonstrates that satisfaction with democracy and trust in political institutions are in decline in Latin America (Booth & Seligson, 2009). This widespread legitimacy crisis has been explained in terms of the gap between citizens’ expectations in Latin American countries and actual performance by the governments in the region (Hagopian, 2005). Several studies have shown that citizens who do not trust political institutions are less likely to engage in conventional political activities (Norris, 2002). In the Latin American context, studies from Bolivia (Smith, 2009), Chile (Carlin, 2006), and Costa Rica (Seligson, 2002) have demonstrated that citizens with higher levels of support for democratic institutions are more likely to vote and to participate in campaign activities. In this article, we assess the impact of satisfaction with democracy on electoral participation, and we expect that more disenchanted voters are less likely to vote.

Another key motivational factor is the perception of electoral integrity. Most Latin American elections are now described as reasonably free and fair by scholars and international observations teams. Undeniably, the formal institutions of procedural democracy have spread in Latin America in the last 30 years (Foweraker & Krznaric, 2002). However, some electoral processes in the region are still marred by a series of irregularities. One of the main problems is that incumbent parties tend to benefit from a disproportionate access to public funds and to broadcast airtime. Moreover, the electoral institutions that have to supervise the elections are in some cases under the control of the executive (Hartlyn, McCoy, & Mustillo, 2008). These irregularities may have a negative impact on citizens’ propensity to vote, because citizens perceive the outcome of the election to be a foregone conclusion (Birch, 2010). Hence, we expect that citizens who have low trust in the quality of elections in Latin America are less likely to vote.

Political information can also affect citizens’ motivation to participate in the elections. Ghirardato and Katz (2002) demonstrate that voters are ambiguity-averse. In other words, they only vote when they are sure that they support the right party, that is, the party that would yield the highest utility to the voter. The main conclusion from this study is that more informed citizens are more likely to vote, because they feel more confident about their electoral choices. Hence, we expect that political information will also reveal itself as an important predictor of electoral participation in Latin America.

Another important motivational factor is political efficacy. The feeling of political efficacy can be described as the perception citizens have of being capable of acting effectively in the political arena. Efficacious citizens perceive that they are capable of influencing government and politics (Craig & Maggiotto, 1982). As a result, they may be more motivated to go to the polls on election day. Recent studies using comparative survey data have indeed shown that more efficacious voters are more likely to vote (Karp & Banducci, 2008; Norris, 2002).

The final variable we consider in this section is party identification. Some studies argue that voting is basically an “expressive” act, and only those citizens who have something to express go to the polls on election day (Achen & Sinnott, 2007; Schuessler, 2000). In the words of Achen and Sinnott (2007, p. 9), “the voters have a sense of acting together with others on behalf of a shared goal, and they derive satisfaction from doing so.” From that perspective, party identification is a key motivational variable. Citizens who are attached to a political party obtain a much higher “expressive” benefit in the elections than those who fail to form political preferences. Moreover, partisanship often works as a “short cut” for voters, helping them to understand political debates and to choose among the different electoral options (Campbell, Converse, Miller, & Stokes, 1960).

Networks of Recruitment

Individual capacity and individual motivation are important determinants of turnout. But the context in which citizens are immersed is also key to understand electoral participation. All other things being equal, citizens who are immersed in networks of political recruitment are more likely to be mobilized to vote.²

Social networks contribute to the mobilization of individuals. Political discussions often occur in nonpolitical institutions of adult life—the working place, voluntary associations, or the church. Hence, these institutions might nurture political interest and increase awareness of the issues at stake in the elections (Verba et al., 1995). Moreover, several studies show that large social networks produce politically relevant social capital (i.e., expertise and political information), which in turn increases the likelihood that citizens will participate in the elections (La Due Lake & Huckfeldt, 1998; Verba et al., 1995). In the American context, church attendance appears to be especially relevant as a factor increasing the political engagement of “unsophisticated” citizens (Alex-Assensoh & Assensoh, 2001; Verba et al., 1995). Kuenzi and Lambright (2011) show that membership in voluntary organizations has a positive impact on electoral participation in African countries. Drawing on data from the 1999-2001 wave of the World Values Survey, Klesner (2007) demonstrates that greater involvement in nonpolitical organizations also leads to more participation in political activities in four Latin American countries (Argentina, Chile, Mexico, and Peru). In line with these previous studies, we expect that Latin American citizens immersed in rich social networks have a higher likelihood of participating in elections. The probability of voting should increase when individuals hold a stable job in the formal economy, join voluntary organizations, and attend church regularly.

Another variable that will be considered in this section is place of residence. Karp, Banducci, and Bowler (2007) contend that the higher population density of cities makes them more attractive locations for parties to canvass. Other scholars argue that the “social pressure” to turn out may be weaker in cities because urbanization tends to produce a “weakening of interpersonal bonds, primary social structures and consensus on norms” (Hoffmann-Martinot, 1994, p. 14). We will include a measure of place of residence (urban vs. rural) in our model below, to assess which of these conflicting theoretical expectations is more applicable to the Latin American context.

The literature on Latin American politics is replete with mentions of “clientelism” and “patronage” (e.g., Calvo & Murillo, 2004; Stokes, 2005). These clientelistic practices appear to be one of the main ailments affecting the quality of democracy in the region. Although the main objective of clientelistic machines is to alter the electoral results in a way that suits the patron, they also work as a tool of electoral mobilization. Even if voters may receive the benefits and vote as they choose, the existence of strong clientelistic networks is likely to increase the incentives for citizens to go to the polls (Nichter, 2008). In fact, it is much easier to supervise whether citizens vote than how they vote. Hence, we expect citizens immersed in clientelistic networks to have a higher probability of voting than the rest of the respondents.

Research Design

Data

Data for the subsequent empirical analysis are drawn from the 2010 Americas Barometer. The survey is administered by the Latin American Public Opinion Project (LAPOP) at Vanderbilt University. The sampling process involves multistage stratification by country and then substratification within each country by major geographic region to increase precision. Within each primary sampling unit (PSU), the survey respondents are selected randomly.³ Honduras drops from the pooled model because one of the items included in our model was not asked in that country (exposure to clientelism), leaving the number of countries analyzed at 17.⁴

Our main variable of interest is electoral turnout. We use a dichotomous measure of the respondents who voted in the last presidential elections: 1 = yes, voted; 0 = no, did not vote. This study focuses on reported turnout rather than actual turnout and privileges reports on past behavior rather than expectations about future voting decisions.⁵

The key independent variables for our study are organized in three groups: capacity, motivation, and networking factors. The first group of variables captures individuals’ capacity to vote. As we mentioned in the theory section, such capacity is determined by the amount of resources available for potential voters. The key determinants of individual capacity to vote are socioeconomic and demographic attributes. The socioeconomic attributes include income and education. The demographic attributes include gender and age. Motivation variables measure individuals’ interest for political issues, their ability to understand what is at stake in the electoral process, and the degree to which they trust the electoral process and the democratic regime. This set of variables includes questions about satisfaction with democracy, trust in elections, political efficacy, interest in politics, party identification, and political awareness. The third group of independent variables seeks to assess the importance of different networks for electoral mobilization. In this case, we evaluate how membership in different social and political organizations shapes individuals’ propensity to vote. We measure respondents’ immersion in different mobilizing networks (voluntary associations and clientelistic networks). We also consider the position of the respondent in the labor market (employment status).⁶

The analysis of the impact of individual motivations and resources on electoral participation definitely enhances our knowledge about electoral behavior in Latin America. However, such knowledge would be incomplete if we do not take into account other contextual variables shaping aggregate turnout at the country level. For this reason, we also include in our analysis important institutional and economic factors that explain aggregate behavior. In line with previous literature, we consider the effect of five institutional and contextual variables: compulsory voting, concurrent elections, closeness of the electoral results, the level of democracy (Polity IV score), and gross domestic product (GDP) per capita (Fornos et al., 2004; Kostadinova & Power, 2007; Pérez-Liñán, 2001).

To examine the effect of compulsory voting on aggregate electoral behavior, we paid special attention to the government’s capacity (or willingness) to enforce the legal mandate. Compulsory voting is only a formal obligation that could be ineffective if it is not accompanied by strong sanctions. Consequently, we created a scale that takes a value of 0 when voting is voluntary, 1 when voting is compulsory but there is no enforcement of this formal rule, and 2 when voting is both mandatory and enforced. The information to create this variable comes from the International Institute for Democracy and Electoral Assistance’s (IDEA) compulsory voting database. We also created a dummy variable to examine the effect of concurrent elections (0 if presidential and legislative elections are not concurrent; 1 if they are concurrent). In addition, our analysis includes a measure of the closeness of the election (i.e., the percentage gap between the first and the second most voted candidates). Finally, we evaluate the effect of country-level economic development on electoral behavior by collecting data on GDP per capita (Penn World Tables).

Model Estimation

Much of what we know about electoral turnout in Latin America is based on the analysis of aggregate-level data. In this article, we use a large number of surveys from diverse national contexts to explore the variation in electoral participation and to test our hypotheses about the specific kinds of motivations, resources, and networks that may explain this variation.

First, we run several logistic regression models to shed light on the effect of motivation, resources, and networks on the decision to vote. The use of logistic regressions is appropriate because our variable of interest—voter turnout—is a dichotomous variable. In the second model, we also include dummy variables for each country to measure whether significant national characteristics (unexplained by the model) lead to different levels of system support.⁷ For these models, we rely on data from the Americas Barometer 2010 for 30,075 respondents in 17 Latin American countries.

Our third model uses multilevel modeling to try to tease out the country-level factors that have an impact on electoral participation. Multilevel models are quickly becoming standard in political science (Snijders & Bosker, 1999) and are usually estimated using either Bayesian simulations or a quasi-likelihood methods (Goldstein, 1995). The most important feature of these models for the purpose of this article is that the estimates of variances and their associated standard errors provide direct tests of the impact of measured contextual effects on turnout. Hierarchical models also allow for a more precise estimation of individual-level factors because they control for important contextual factors that may bias the results.

We use a mixed-effects model for binary responses because the grouping structure of the data consists of two levels of nested groups (individuals nested in countries). The first level of the model describes the distribution of the individual observations—that we assume are normally distributed—and transforms the model-based predicted values. First-level predictors are grouped in three sets: resources, motivations, and networks. Second-level predictors are country-level predictors that explain aggregate electoral behavior. In particular, we estimate a random coefficient model. In this type of models, the coefficients (or slopes) vary across clusters, and we estimate separate income, age, and education differences for each country. We assume that the differences are drawn from a normal distribution and that the covariance matrix of the random effects is unstructured, that is, we allow correlation between the level of turnout and the differences in income, age, and education in each country.

Results

Table 1 presents three models estimating voter turnout at the individual level of analysis. Model 1 estimates a logistic model for the effects of motivations, resources, and networks on voting. In this first model, we clustered the standard errors by country. Model 2 is a logistic regression with country fixed effects. Model 3 estimates a multilevel random coefficient model.

Table 1.

Determinants of Voter Turnout in Latin America, Logistic, Fixed Effects, and Multilevel Logistic Models.

	(1) Logistic	(2) Fixed effects	(3) Random coefficient
	b/(SE)	b/(SE)	b/(SE)
Individual-level variables
Resources
Income	0.007 (0.03)	0.012 (0.01)	0.013 (0.01)
Education	0.367*** (0.06)	0.306*** (0.03)	0.306*** (0.03)
Age	0.665*** (0.06)	0.675*** (0.02)	0.675*** (0.02)
Gender (male=1)	−0.233*** (0.06)	−0.209*** (0.04)	−0.208*** (0.04)
Motivations
Satisfaction	0.070 (0.04)	0.026 (0.03)	0.026 (0.03)
Trust elections	0.088*** (0.02)	0.056*** (0.01)	0.057*** (0.01)
Political efficacy	−0.004 (0.02)	−0.026** (0.01)	−0.026** (0.01)
Interest in politics	0.105* (0.04)	0.134*** (0.02)	0.133*** (0.02)
Political awareness	0.091** (0.03)	0.114*** (0.02)	0.113*** (0.02)
Partisanship	0.377*** (0.11)	0.553*** (0.05)	0.555*** (0.05)
Networks
Employment status	0.500*** (0.05)	0.532*** (0.04)	0.530*** (0.04)
Church attendance	−0.026 (0.03)	−0.001 (0.02)	−0.002 (0.02)
Membership index	0.136** (0.04)	0.180*** (0.02)	0.181*** (0.02)
Urban/Rural	−0.102 (0.11)	−0.115* (0.05)	−0.111* (0.05)
Clientelism	0.062 (0.06)	0.102** (0.04)	0.102** (0.04)
Country-level variables
Compulsory vote			0.491* (0.21)
Concurrent elections			0.910 (0.48)
Closeness			0.021* (0.01)
GDP per capita			−0.000 (0.00)
Polity IV			−0.106 (0.10)
Random effects
SD (age)			0.255*** (0.05)
SD (education)			0.131*** (0.04)
SD (income)			0.026*** (0.01)
N individuals	22,457	22,457	22,457
N countries	17	17	17

p < .1, two-tailed.**p < .05, two-tailed. ***p < .001, two-tailed.

The findings regarding the effect of resources on individual voting behavior are revealing. More educated individuals are more prone to participate in electoral processes. In fact, the likelihood that individuals will vote in presidential elections significantly increases as they become more educated. Our results also offer convincing evidence in support of the argument that older individuals are more likely to vote than younger individuals. The coefficient for the variable age is positive and statistically significant in all the models presented in Table 1.

At first glance, all the models in Table 1 suggest that income is not a significant predictor of electoral participation in Latin America. Contrary to the expectations of the SES model of voting behavior, wealthy individuals do not necessarily vote more than poor individuals. However, we tried a different model specification that revealed that the relationship between personal wealth and electoral participation exists, but it is not linear. We run the same models presented above, but this time including all the income categories as dummies, excluding the highest and the lowest income categories that served as baseline categories (see Table 2).⁸

Table 2.

Logistic Regressions With Income Dummies.

Income level	Full model with income dummies (baseline 10th decile)	Full model with income dummies (baseline no income)
No income	−0.463** (0.190)
1st decile	−0.119 (0.164)	0.344*** (0.126)
2nd decile	−0.003 (0.159)	0.460*** (0.124)
3rd decile	−0.027 (0.157)	0.436*** (0.123)
4th decile	0.053 (0.157)	0.516*** (0.124)
5th decile	−0.044 (0.158)	0.420*** (0.127)
6th decile	−0.114 (0.160)	0.350*** (0.132)
7th decile	0.065 (0.167)	0.528*** (0.141)
8th decile	−0.040 (0.177)	0.423*** (0.153)
9th decile	−0.157 (0.189)	0.307* (0.168)
10th decile		0.463** (0.189)

p < .1, two-tailed. **p < .05, two-tailed. ***p < .001, two-tailed.

The results suggest that those who have no income at all tend to vote less than all the other respondents. In sum, the level of income does not matter, but destitute individuals vote less than the rest of the population.⁹ Because extreme poverty is more common in some Latin American countries than in the industrialized world, this is an interesting finding that is in line with our theoretical expectation.¹⁰

The impact of gender on voter turnout is also worth noting. The few studies that analyze women’s political involvement in Latin America conclude that there is a “gender gap” in political participation and that women are less likely to be politically engaged in the region (Desposato & Norrander, 2009). Our results, however, show the opposite effect. Women appear to vote more than men in Latin American elections.¹¹ Desposato and Norrander (2009) point out that the “gender gap” in political participation is attenuated when there is a high level of women participation in public office—which is an effect of “symbolic representation.”¹² In a similar vein, our findings about women’s higher turnout in Latin America may be capturing important changes to women’s status in the political arena that are causing women to vote more, thus challenging the conventional wisdom. For instance, due to gender quotas, women have very high levels of parliamentary representation in Argentina, Mexico, Peru, and the Dominican Republic; and Uruguay recently passed a gender quota law that will boost its numbers (Jennifer Piscopo, personal communication, August 1, 2012). As can be observed in Table 3, these are precisely the countries where women tend to vote more than men.

Table 3.

Voter Turnout Latin America, by Country.

	ARG	BOL	BRA	CHI	COL	CR	DR	ECU	GUA	HON	MEX	NIC	PAN	PAR	PER	SAL	URU	VEN
Resources
Income	0.09*	0.06	−0.10**	−0.03	0.05	0.06	0.03	−0.03	−0.01	−0.09**	−0.02	0.06	−0.05	0.00	0.04	−0.02	0.04	0.10**
Education	0.16	0.34**	0.41**	0.13	0.04	0.25*	0.22*	0.39**	0.45***	0.28**	0.44***	0.12	0.26*	0.34**	0.90***	0.41***	0.36*	0.27**
Age	0.31***	0.56***	0.80***	0.08	0.99***	0.91***	0.65***	0.12*	0.78***	0.20**	0.93***	0.50***	0.66***	0.90***	1.25***	0.50***	0.73***	0.63***
Gender (male=1)	−0.55**	0.04	−0.20	−0.47	−0.25*	−0.09	−0.41**	−0.26	0.33**	0.17	−0.67***	−0.19	−0.28	0.14	−0.35*	−0.19	−0.51*	−0.12
Motivations
Satisfaction democracy	0.15	−0.16	0.12	−0.22	0.01	0.23**	0.09	−0.12	0.03	0.11	0.04	−0.15	0.02	0.03	−0.25*	−0.01	0.24	0.09
Trust elections	−0.02	0.06	0.04	0.13	0.13**	0.04	−0.01	−0.01	0.04	0.13**	0.10**	0.06	0.10*	0.03	0.01	0.09**	0.19**	0.10**
Political efficacy	−0.07	0.07	−0.02	−0.07	−0.02	−0.06	−0.03	0.05	−0.08**	0.06	−0.02	0.01	−0.12**	0.03	0.08	−0.04	−0.09	−0.05
Interest in politics	0.25**	0.26**	0.05	0.38**	0.23**	0.18**	0.27**	−0.06	0.10	0.43***	0.16**	0.24**	0.01	0.32***	−0.14	0.14*	−0.01	0.02
Political awareness	0.21**	0.13	0.09	0.17	0.11	0.18*	−0.01	0.17*	0.09	0.19**	0.19**	0.07	0.09	0.27*	−0.02	0.09	−0.28	0.16**
Partisanship	0.34	0.11	0.33**	0.75	0.28*	1.23***	0.75***	0.09	0.36*	1.06***	0.50**	0.68***	1.55***	0.61***	0.17	0.38**	0.10	0.74***
Networks
Employment	0.40**	0.47**	0.78***	0.61**	0.63***	0.11	0.37**	0.78***	0.84***	.35**	0.35**	0.19	0.71***	0.54**	1.04***	0.37**	0.76**	0.56***
Church attendance	−0.01	0.03	−0.01	0.24**	−0.02	−0.00	−0.07	0.07	0.11*	.02	−0.02	0.02	0.07	0.02	−0.17**	−0.01	−0.04	−0.11*
Membership	−0.18	0.04	0.22**	0.02	0.30**	0.03	0.18**	0.07	0.27**	.02	0.21**	0.20**	0.06	0.05	0.30**	0.41***	0.53**	0.29***
Urban/rural	0.10	0.01	−0.29	−1.62**	−0.24	−0.08	−0.31	−0.11	−0.06	−0.11	−0.03	−0.34**	−0.11	0.16	−0.32	0.24	0.46	0.25
Clientelism	0.16	0.45**	−0.00	−0.54	0.30**	0.03	0.46***	0.08	0.14	NA	0.07	−0.03	0.03	0.28**	0.23	−0.23	0.19	−0.22
Constant	−1.74**	−1.63*	−1.74**	1.83	−4.09***	−5.11***	−1.83**	0.65	−3.66***	0.13***	−3.70***	−1.32*	−1.58*	−5.09***	−2.65***	−2.09***	−0.79	−3.59***
n	886	2,121	1,892	1,083	1,196	986	1,164	2,336	1,104	1,255	1,257	1,178	1,245	971	1,190	1,423	1,279	1,146

NA = question not asked in this country.

p < .1, two-tailed. **p < .05, two-tailed. ***p < .001, two-tailed.

Looking at the coefficients for motivational variables, our findings underline the importance of trust in elections, interest in politics, partisanship, and political awareness as significant predictors of voting turnout. All these variables have significant and positive effects on electoral participation. That is, trustful, interested, partisan, and informed citizens are more likely to participate in elections. In contrast, satisfaction with democracy has no effect on individuals’ decision to vote. Surprisingly, the coefficient for political efficacy is significant but negative in Models 2 and 3. In other words, the perception that government is responsive to individuals’ concerns seems to discourage electoral participation. This counterintuitive finding may simply reflect that voting offers a possibility for disenchanted voters who feel unefficacious to express their frustration by supporting antisystemic or outsider candidates (Carreras, 2012; Doyle, 2011). Citizens who perceive that the government pays attention to their aspirations may paradoxically become more apathetic. However, the impact of efficacy on turnout is very small as it will be revealed in the next section.¹³

Finally, we also evaluate the effect of recruitment networks on voters’ mobilization. Table 1 shows that individuals who are employed and citizens actively engaged in civic associations are significantly more likely to vote than unemployed individuals and unengaged individuals. As we hypothesized, citizens involved in large social networks are more likely to be politically engaged and to participate in electoral processes.

The results also suggest that individuals’ place of residence influences their electoral behavior. Models 2 and 3 present evidence that the likelihood that an individual votes in presidential elections decreases if the voter lives in urban areas. This finding may be explained by the greater social cohesion in rural areas. Social pressure to participate may be felt much more strongly in rural areas, which house more tightly knit communities (Bratton, Chu, & Lagos, 2010; Hoffmann-Martinot, 1994). Moreover, conditions for political mobilization through patronage networks are more favorable in rural areas, as a recent study of electoral participation in Africa has demonstrated (Kuenzi & Lambright, 2011).

In sum, Table 1 shows the main factors explaining individual incentives to vote in Latin America. As discussed in the theory section, these incentives are driven by political motivations, individual sociodemographic attributes, and social networks. Older, educated, politically motivated, and civic-active citizens are more likely to vote in Latin America.

Models 2 and 3 also evaluate the effect of contextual-level and institutional variables. For ease of presentation, the estimates of the 17 country dummies in Model 2 are not shown. In all but two cases, the coefficients of the country dummy variables were statistically significant at p < .1 in a two-tailed test. The sign and magnitude of the specific country coefficients are not, in and of themselves, of interest here, but the results suggest that it is important to take contextual and institutional factors into account when explaining electoral participation in the region. To tease out some of these contextual factors, Model 3 estimates a multilevel logistic model. This final model also evaluates how institutional, political, and economic conditions at the country level could change the effect of our variables of interest on electoral turnout. Model 3 estimates such effects using a two-level random-coefficients model.

Five contextual variables are incorporated into our final model: compulsory voting, concurrent elections, closeness of the electoral result (percentage gap between the first and the second most voted candidates), the Polity IV score (level of democracy), and GDP per capita. In line with previous literature (Fornos et al., 2004; Kostadinova & Power, 2007), the existence of compulsory voting laws effectively shapes individual behavior. The likelihood that an individual decides to vote in presidential elections increases significantly in countries where voting is compulsory, especially in countries where such laws are effectively enforced. Electoral participation also tends to be higher when the elections are competitive, that is, when the difference between the first and the second most voted candidates in the presidential elections is small (see Cox & Munger, 1989). However, our results cast doubts on previous findings regarding the link between turnout and concurrent elections. As expected, the coefficient for the variable measuring whether the elections are concurrent is positive, but it does not reach statistical significance. Similarly, the level of democracy (Polity IV score) and GDP per capita are not significant predictors of electoral participation in Latin American elections. Overall, our results confirm most of the findings of previous studies regarding the contextual factors that influence electoral participation in Latin America (Fornos et al., 2004; Kostadinova & Power, 2007; Pérez-Liñán, 2001).

Which Theoretical Perspective Explains More in Latin America?

The previous section showed that a series of variables are statistically significant individual-level predictors of turnout in Latin America. Our main objective, however, is to assess which of the three perspectives detailed above (resources, motivations, and immersion in mobilization networks) carries more explanatory power. To estimate precisely what effect the independent variables of interest have on the probability of voting, we calculated the predicted probabilities of participating in the elections. Table 4 presents the predicted probabilities of voting at different values of the independent variables that were statistically significant in Model 3 above, holding all other variables at their means. As an additional way of estimating more precisely the impact of specific variables on the likelihood of electoral participation in Latin America, we run the logistic regression in Table 1 for each individual country. This allows us to evaluate whether the statistically significant variables in the pooled model are also significant predictors of turnout in individual Latin American countries. We present this information in Table 3.

Table 4.

Predicted Probabilities of Voting in Latin America (2010).

Value on the independent variables	Predicted probability
Resources
No education	.77
Primary education	.82
Secondary education	.86
Higher education	.89
Age 18-24	.66
Age 25-34	.79
Age 35-49	.88
Age 50-64	.94
Age > 64	.97
Partisan	.83
Not partisan	.90
Women	.87
Men	.84
Motivations
Low trust in elections (1)	.83
High trust in elections (7)	.87
Low efficacy (1)	.86
High efficacy (7)	.85
Low political interest (1)	.84
High political interest (4)	.88
Low political information (1)	.80
High political information (5)	.86
Mobilization networks
Unemployed	.81
Employed	.88
No membership civic organizations	.83
Membership 1 civic organization	.86
Membership 2 civic organizations	.88
Membership +3 civic organizations	.90
Rural	.87
Urban	.85
Never exposed to clientelism	.85
Sometimes exposed to clientelism	.86
Often exposed to clientelism	.87
Contextual factors
No compulsory voting	.76
Compulsory voting, but not enforced	.84
Compulsory voting, enforced	.90
Closeness (minimum)	.80
Closeness (maximum)	.89

The predicted probabilities clearly show that two “resources” variables (age and education) stand out as the best individual-level predictors of electoral participation in our model. The table demonstrates that a strong socialization effect exists. Age can be considered as a proxy for political experience. As expected from the socialization hypothesis, the youngest voters (age 18-24) are much less likely to vote than the rest of the population. This suggests that voters get socialized into politics quite fast in the workplace or in their social networks. The likelihood of voting keeps increasing as age increases, but the gap between the different age categories gradually diminishes. This is consistent with the “start-up-slow-down-model” developed by Verba and Nie (1972).¹⁴ Age is not only the strongest predictor of turnout according to the predicted probabilities, but it is also the most consistent one across the region (see Table 3).¹⁵

The second strong predictor of electoral participation in Latin America is education. We hypothesized that turnout should increase as the level of education increases, because educated individuals are more likely to absorb complex political information and become politically interested. This is exactly what the predicted probabilities reveal. Although the probability of voting keeps increasing as the level of education increases, the biggest differences are the ones that exist between no education versus primary, and primary versus secondary. As with the variable age, education is also a statistically significant predictor of electoral participation in the vast majority of Latin American countries (14 of 18).

The motivation variables also have an impact on the probability of participating in the elections, but the substantive effect of these factors pales in comparison with the variables “age” and “education.” Three variables (trust in elections, political information, and political interest) have a moderate impact on electoral participation. As can be observed in Table 4, the gap between those who express low and high values for these three variables is between 4% and 6%, which is a significant impact but substantively less important than the effect of voters’ resources. Similarly, these three variables are statistically significant predictors of turnout in a considerable number of countries (7 countries for the variables “trust in elections” and “political information,” 10 countries for the variable “political interest”), but they do not explain electoral participation in many other nations.

Two variables stand out in the set of motivation predictors (political efficacy and partisanship), but for different reasons. In the previous section, we showed that political efficacy is a statistically significant predictor of turnout, but in the unexpected direction (more efficacious citizens appear to vote less). However, the analysis of the predicted probabilities and of the country-level regressions reveals that the feeling of efficacy has very little influence on electoral participation. Citizens with low efficacy are only 1% more likely to vote than individuals with high efficacy, and efficacy is a statistically significant predictor of turnout in only two countries. On the contrary, partisanship stands out as the strongest predictor of electoral participation among the motivation variables (difference of seven percentage points in the probability of voting between partisans and nonpartisans). Moreover, this effect is very consistent across countries (see Table 3).

The analysis of the network variables reveals a nuanced picture. While some factors appear to be relatively strong predictors of electoral participation (membership in civic organizations, employment status), other variables do not explain much of the variance in citizens’ decision of whether or not to go to the polls on election day (clientelism, church attendance, place of residence).

Let us consider the weak predictors first. The pooled model suggested that church attendance has very little impact on turnout. Table 3 confirms this finding, as this variable is a statistically significant predictor of turnout in four countries only. Place of residence is also a weak predictor of electoral participation in Latin America. The predicted probabilities reveal that urban residents are only slightly less likely to vote than rural residents (87% vs. 85%). Moreover, Table 3 shows that the place of residence is a statistically significant predictor of electoral participation in three Latin American countries only. Clientelism is another surprisingly weak predictor of electoral participation in Latin America. Although political clientelism is a statistically significant predictor of turnout in the pooled multilevel model, the predicted probabilities of voting increase only slightly when citizens are exposed to clientelism in their networks. Moreover, Table 3 reveals that this factor is a statistically significant predictor of turnout in only four countries.

The other mobilization variables in our model have a much stronger impact on electoral participation. Full-time employment and membership in civic organizations are statistically significant predictors of electoral participation in the pooled hierarchical model (Table 1) and in the majority of countries (Table 3). In addition, the predicted probabilities in Table 4 demonstrate a strong substantive impact of these two “network” variables on turnout.

It is useful to compare the predicted probabilities of electoral participation presented so far with the predicted probabilities of voting at different levels of the contextual factors. Conclusions about the impact of contextual factors on citizens’ decision to vote are more tentative because the number of observations in the second level of analysis is relatively low (17). The two contextual factors that are statistically significant in the hierarchical models appear to have a strong impact on electoral participation. Individuals living in countries that have compulsory voting laws are much more likely to go to the polls on election day. Similarly, individuals have a significantly higher probability of voting when they face close electoral contests. These effects are similar in size to the impact of the most influential individual-level factors.¹⁶

As a final probe of the substantive impact of each theoretical perspective, we calculated measures of fit for the full logistic model and for reduced models excluding the variables from one or two of the three perspectives. The McKelvey & Zavoina’s R² and McFadden’s R² for these reduced models are presented in Table 5. These measures of fit confirm our overall findings. Both measures of fit produce significantly higher values in the Reduced Model 1 (including only the set of resources variables) than in the Reduced Model 2 (motivations model), and in the Reduced Model 3 (networks model). We reach the same conclusion if we compare the change in R² attributable to each of the three set of variables through entering them to the equation as the last step. Comparing the Reduced Models 4, 5, and 6 with the full model, it is clear that entering the set of resource variables as the last step increases the variability explained much more than if the set of motivation variables or mobilization variables are entered to the equation as the final step. In sum, all our results point to the importance of voters’ resources—especially age and education—in explaining electoral participation.

Table 5.

Comparing Models of Electoral Turnout in Latin America (Measures of Fit).

	McKelvey & Zavoina’s R²	McFadden’s R²
Reduced Model 1: Resources	.174	.092
Reduced Model 2: Motivations	.034	.019
Reduced Model 3: Networks	.036	.019
Reduced Model 4: Resources + Motivations	.196	.103
Reduced Model 5: Resources + Networks	.191	.104
Reduced Model 6: Motivations + Networks	.062	.034
Full model	.212	.116

The importance of voter’s resources to explain turnout in Latin America contrasts with the little influence that variables such as income or education have on electoral participation in developed countries. Particularly, education is a very poor predictor of electoral participation in many industrialized countries (Nevitte, Blais, Gidengil, & Nadeau, 2009; Verba et al., 1978). Why are citizens with a low SES (i.e., destitute and poorly educated individuals) less likely to go to the polls in Latin America but not in most industrialized countries? We believe there are two main reasons that explain this pattern. First, the gap between those that have a low level of education and those that have a high level of education is more remarkable in Latin America than in most industrialized countries. As we showed in our analysis, the citizens that are least likely to vote are those who did not complete primary education (34.5% of the respondents in our sample find themselves in this situation). Since most citizens in developed countries crossed this minimum threshold of instruction (the vast majority of citizens at least completed primary school), it makes sense that the effect of education on electoral participation is less remarkable.¹⁷ Second, the literature suggests that voters’ resources will matter less when leftist parties or labor movements are able to mobilize lower status individuals (Gallego, 2010). Latin American countries have lacked precisely the type of labor parties that were created in Europe in the 20th century to mobilize the working-class electorate (Bartolini, 2000). Latin American party systems have traditionally been dominated by “parties of a multiclass appeal and ideological pragmatism” (Dix, 1989, p. 33). These catch-all parties do not develop programmatic linkages with voters along existing lines of societal cleavages and are less effective at mobilizing individuals with low SES. Moreover, the neoliberal turn in the 1990s has considerably weakened labor movements in the region, thereby eroding a potential mobilization arena that could encourage disadvantaged social groups to go to the polls (Roberts, 2002). In sum, a series of structural factors help explain the divergent impact of voters’ resources on electoral participation across different regions.

Conclusion

In this article, we assessed the relevance of three different theoretical perspectives (resources, motivations, and mobilization networks) to explain electoral participation in Latin America. The empirical results do not show one of the three theoretical perspectives as uniformly better at explaining turnout in the region. Within each perspective, some variables are strong predictors of electoral participation, while other factors only marginally influence turnout. But the strongest predictors in all of our models are two individual resources (education and age—proxy for political experience). Our analysis reveals that these objective characteristics of the voters explain much more than their subjective motivations (trust in elections, political efficacy, and interest in politics). We also show that some mobilization networks are important to activate turnout in Latin America (civic associations, working place), but other networks do not affect electoral participation in the region—or do it only marginally (place of residence, clientelistic networks, religious congregations).

The conventional wisdom regarding turnout in Latin America is that institutions matter much more than socioeconomic factors. In the most comprehensive analysis of electoral participation in the region to date, Fornos et al. (2004) indeed conclude that “socioeconomic variables, which are found to have strong effects on turnout in Western democracies, are unrelated to turnout in Latin American countries” (p. 909). The present analysis challenges the accepted wisdom. We demonstrate that the strongest predictors of turnout in the region (education, age, employment status) are all socioeconomic variables. Income also matters, but the impact is not linear. Our analysis reveals that individuals in situation of extreme poverty are less likely to vote than the rest of the population.

Footnotes

Acknowledgements

We thank two anonymous reviewers, Steven Finkel, Aníbal Pérez-Liñán, Scott Morgenstern, Ignacio Arana, Cassilde Schwartz, Jennifer Piscopo, Ryan Carlin, and Markus Steinbrecher for useful comments and suggestions on previous drafts of this article.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Miguel Carreras is a PhD candidate in political science at the University of Pittsburgh. He holds an MA from the Graduate Institute of International Studies (Geneva). His current research focuses on presidential elections and party system change in Latin America after the Third Wave of democratization.

Néstor Castañeda-Angarita is a PhD candidate in the Department of Political Science at the University of Pittsburgh. He earned a BA in economics from the Universidad Nacional de Colombia and an MA in Latin American studies from the University of Texas at Austin. His research interests include comparative political economics, legislative politics, and formal theory. His doctoral dissertation addresses the influence of special interest groups on fiscal politics and deficit reduction in Latin America.

References

Achen

C. H.

Sinnott

(2007, April). Taking sides: Learning and voting. Paper presented at the Annual Meeting of the Midwest Political Science Association, Chicago, IL.

Alex-Assensoh

Assensoh

A. B.

(2001). Inner-city contexts, church attendance, and African-American political participation. Journal of Politics, 63, 886-901.

Bartolini

(2000). The political mobilization of the European Left, 1860-1980: The class cleavage. New York, NY: Cambridge University Press.

Birch

(2010). Perceptions of electoral fairness and voter turnout. Comparative Political Studies, 43, 1601-1622.

Blais

(2007). Turnout in elections. In Dalton

R. J.

Klingemann

H.-D.

(Eds.), The Oxford handbook of political behavior (pp. 621-635). New York, NY: Oxford University Press.

Booth

J. A.

Seligson

M. A.

(2009). The sources of political legitimacy. In The legitimacy puzzle: Political support and democracy in Latin America (pp. 105-143). New York, NY: Cambridge University Press.

Brady

H. E.

Verba

Schlozman

K. L.

(1995). Beyond SES: A resource model of political participation. American Political Science Review, 89, 271-294.

Bratton

Chu

Y.-h.

Lagos

(2010). Who votes? Implications for new democracies. Taiwan Journal of Democracy, 6, 107-136.

Calvo

Murillo

M. V.

(2004). Who delivers? Partisan clients in the Argentine electoral market. American Journal of Political Science, 48, 742-757.

10.

Campbell

Converse

P. E.

Miller

W. E.

Stokes

D. E.

(1960). The American voter. New York, NY: Wiley.

11.

Carlin

R. E.

(2006). The decline of citizen participation in electoral politics in post-authoritarian Chile. Democratization, 13, 632-651.

12.

Carreras

(2012). The rise of outsiders in Latin America, 1980-2010: An institutionalist perspective. Comparative Political Studies, 45, 1451-1482.

13.

Cox

G. W.

Munger

M. C.

(1989). Closeness, expenditures, and turnout in the 1982 US house elections. American Political Science Review, 83, 217-231.

14.

Craig

S. C.

Maggiotto

M. A.

(1982). Measuring political efficacy. Political Methodology, 8, 85-109.

15.

Desposato

Norrander

(2009). The gender gap in Latin America: Contextual and individual influences on gender and political participation. British Journal of Political Science, 39, 141-162.

16.

Dix

R. H.

(1989). Cleavage structures and party systems in Latin America. Comparative Politics, 22, 23-37.

17.

Doyle

(2011). The legitimacy of political institutions: Explaining contemporary populism in Latin America. Comparative Political Studies, 44, 1447-1473.

18.

Fornos

C. A.

Power

T. J.

Garand

J. C.

(2004). Explaining voter turnout in Latin America, 1980 to 2000. Comparative Political Studies, 37, 909-940.

19.

Foweraker

Krznaric

(2002). The uneven performance of third wave democracies: Electoral politics and the imperfect rule of law in Latin America. Latin American Politics and Society, 44, 29-60.

20.

Gallego

(2010). Understanding unequal turnout: Education and voting in comparative perspective. Electoral Studies, 29, 239-248.

21.

Ghirardato

Katz

J. N.

(2002). Indecision theory: Quality of information and voting behavior (Social Science Working Paper 1106R). Pasadena: California Institute of Technology.

22.

Goldstein

(1995). Multilevel statistical models. New York, NY: Oxford University Press.

23.

Hagopian

(2005). Conclusions: Government performance, political representation, and public perceptions of contemporary democracy in Latin America. In Hagopian

Mainwaring

S. P.

(Eds.), The third wave of democratization in Latin America: Advances and setbacks (pp. 319-362). New York, NY: Cambridge University Press.

24.

Hartlyn

McCoy

Mustillo

T. M.

(2008). Electoral governance matters: Explaining the quality of elections in contemporary Latin America. Comparative Political Studies, 41, 73.

25.

Hoffmann-Martinot

(1994). Voter turnout in French municipal elections. In Lopez-Nieto

(Ed.), Local elections in Europe (pp. 13-42). Barcelona, Spain: Institut de ciènces politiques I socials.

26.

Karp

J. A.

Banducci

S. A.

(2008). Political efficacy and participation in twenty-seven democracies: How electoral systems shape political behaviour. British Journal of Political Science, 38, 311-334.

27.

Karp

J. A.

Banducci

S. A.

Bowler

(2007). Getting out the vote: Party mobilization in a comparative perspective. British Journal of Political Science, 38, 91-112.

28.

Karp

J. A.

Brockington

(2005). Social desirability and response validity: A comparative analysis of overreporting voter turnout in five countries. Journal of Politics, 67, 825-840.

29.

Klesner

J. L.

(2007). Social capital and political participation in Latin America: Evidence from Argentina, Chile, Mexico, and Peru. Latin American Research Review, 42, 1-32.

30.

Kostadinova

Power

T. J.

(2007). Does democratization depress participation? Voter turnout in the Latin American and Eastern European transitional democracies. Political Research Quarterly, 60, 363.

31.

Kuenzi

Lambright

G. M. S.

(2011). Who votes in Africa? An examination of electoral participation in 10 African countries. Party Politics, 17, 767-799.

32.

La Due Lake

Huckfeldt

(1998). Social capital, social networks, and political participation. Political Psychology, 19, 567-584.

33.

Lehoucq

Wall

D. L.

(2004). Explaining voter turnout rates in new democracies: Guatemala. Electoral Studies, 23, 485-500.

34.

Leighley

J. E.

Nagler

(1992). Individual and systemic influences on turnout: Who votes? 1984. Journal of Politics, 54, 718-740.

35.

Lijphart

(1997). Unequal participation: Democracy’s unresolved dilemma. American Political Science Review, 91, 1-14.

36.

Nevitte

Blais

Gidengil

Nadeau

(2009). Socioeconomic status and nonvoting: A cross-national comparative analysis. In Klingemann

H.-D.

(Ed.), The comparative study of electoral systems (pp. 109-136). New York, NY: Oxford University Press.

37.

Nichter

(2008). Vote buying or turnout buying? Machine politics and the secret ballot. American Political Science Review, 102, 19-31.

38.

Niemi

R. G.

Barkan

J. D.

(1987). Age and turnout in new electorates and peasant societies. American Political Science Review, 81, 583-588.

39.

Norris

(2002). Democratic phoenix: Reinventing political activism. New York, NY: Cambridge University Press.

40.

Pateman

(1976). Participation and democratic theory. New York, NY: Cambridge University Press.

41.

Pérez-Liñán

(2001). Neoinstitutional accounts of voter turnout: Moving beyond industrial democracies. Electoral Studies, 20, 281-297.

42.

Pitkin

H. F.

Shumer

S. M.

(1982). On participation. Democracy, 2, 48-53.

43.

Remmer

K. L.

(2010). Political scale and electoral turnout: Evidence from the less industrialized world. Comparative Political Studies, 43, 275-303.

44.

Roberts

K. M.

(2002). Social inequalities without class cleavages in Latin America’s neoliberal era. Studies in Comparative International Development, 36, 3-33.

45.

Rosenstone

S. J.

Hansen

J. M.

(1993). Mobilization, participation and democracy in America. New York, NY: Longman.

46.

Schuessler

A. A.

(2000). A logic of expressive choice. Princeton, NJ: Princeton University Press.

47.

Schwindt-Bayer

L. A.

(2010). Political power and women’s representation in Latin America. New York, NY: Oxford University Press.

48.

Seligson

M. A.

(2002). Trouble in paradise? The erosion of system support in Costa Rica, 1978-1999. Latin American Research Review, 37, 160-185.

49.

Smith

A. E.

(2009). Legitimate grievances: Preferences for democracy, system support, and political participation in Bolivia. Latin American Research Review, 44, 102-126.

50.

Snijders

T. A.

Bosker

R. J.

(1999). Multilevel analysis: An introduction to basic and advanced multilevel modeling. London, England: Sage.

51.

Stokes

S. C.

(2005). Perverse accountability: A formal model of machine politics with evidence from Argentina. American Political Science Review, 99, 315-325.

52.

Verba

Nie

N. H.

(1972). Participation in America: Political democracy and social equality. Chicago, IL: The University of Chicago Press.

53.

Verba

Nie

N. H.

Kim

J.-O.

(1978). Participation and political equality: A seven-nation comparison. Chicago, IL: University of Chicago Press.

54.

Verba

Schlozman

K. L.

Brady

H. E.

(1995). Voice and equality: Civic voluntarism in American politics. Cambridge, MA: Harvard University Press.

55.

Wolfinger

R. E.

Rosenstone

S. J.

(1980). Who votes? New Haven, CT: Yale University Press.