Socialization and Political Regimes: the Impact of Generation on Support for Democracy in Latin America*

Studies on democratic attitudes in Latin America indicate that older citizens are more likely to express a preference for democracy. This contradicts part of the literature, which suggests that the greatest support should come from younger generations, who were socialized under democratic regimes. One possible explanation for the greater support for democracy among the older generation is that they experienced the repression of political and civil rights under authoritarian rule, thus creating an aversion to such regimes. In this article, we replicate tests conducted by other studies in evaluating the effect of generation on support for democracy, using data from the 2012 Americas Barometer. In addition, we add a new factor to the analysis: the country's authoritarian legacy, measured as to the duration and intensity with which individual and political rights were curtailed in the past. The results show a complex picture. First, they confirm that the generations that have lived under authoritarian regimes are more likely to support democracy. However, we find no evidence of an increase in the difference in support between generations in countries where the authoritarian legacy is stronger. Finally, the data indicate that countries with stronger authoritarian legacies exhibit less support for democracy, while stronger democratic legacies have the opposite effect.

(2018) 12 (1) e0001 -2/22 major, recurring theme in political socialization studies is the relationship between the experiences of different generations and their political attitudes and behavior (INGLEHART and WELZEL, 2005;JENNINGS and NIEMI, 1981;MANNHEIM, 1952;NEUNDORF and NIEMI, 2014). However, most studies in the field have been based on empirical research conducted in countries with long democratic traditions, which precludes the comparison of generations that have lived under different political regimes. Responding to this lacuna, recent studies have turned their attention to the role of the generational differences in the context of new democracies, such as the countries from the former USSR (MISHLER and ROSE, 1999, Asia (CHU et al., 2008) and Africa (BRATTON, MATTES and GYIMAH-BOADI, 2005).
Studies conducted in Latin America have also considered the generational question, focusing on its effects on the formation of attitudes (MORENO and LAGOS, 2016) and, more broadly, on the development of civic and democratic political culture (BAQUERO, 2004).
Are there differences in democratic attitudes between generations that have only lived under democratic regimes and those that have also experienced authoritarianism? If such differences exist, in which direction do they tend to go? Do younger cohorts, socialized only under democracy, have more democratic attitudes? Or, on the contrary, is it older cohorts, who also lived under authoritarian regimes, that are more attached to democracy? These are some of the questions raised by studies on generations and democratic legitimacy in the context of new democracies.
Opinion is divided over how to answer them. The study by Mishler and Rose (2007), for example, shows that there are small but significant differences between generations in Russia, with those cohorts who lived under the communist regime showing a certain nostalgia for it and less satisfaction with democracy. In studies conducted in Asia and Africa, respectively, Chu et al. (2008) and Bratton et al. (2005) find few differences between generations, and little change in this over time.
In Latin America, few studies have investigated generational effects on support for democracy. One exception to this is a study by Moreno and Lagos (2016), which tests the hypothesis that individuals socialized under democratic regimes have more democratic attitudes. In general, their results confirmed this hypothesis. However, they also found that cohorts socialized under democratic regimes that existed prior to the authoritarian period displayed a greater commitment to democracy than cohorts that A (2018) 12 (1) e0001 -3/22 have spent their formative years during the most recent democratic period. One of the explanations suggested by the authors is that, because they have experienced the 'horrors' of the military rule, those among the older democratic cohort are more likely to recognize the virtues of democracy.
In this article, we return to this explanation, which we will call the 'argument of aversion to authoritarian rule'. However, unlike Moreno and Lagos (2016), our aim is not to compare generations who spent their formative years under different political regimes, but rather to compare those generations that were socialized only under democracy with those that also experienced authoritarianism. By comparing these two groups, it is possible to answer our two research questions: 01. does experience of authoritarianism and, therefore, the ability to compare it with democracy, increase support for democracy? 02. is this effect greater in contexts where authoritarian regimes were harsher and more repressive?
In order to answer both questions, we used the data from 17 countries of the there is more support for democracy, while in countries with stronger authoritarian legacies, support is weaker.
The article is structured in four sections. The first and second sections present, respectively, the key debates on political socialization and generations, and on generational differences and support for democracy. In the third section, we explain how we constructed the main variables used in the study, including the division of 'generations', and the 'democratic' and 'authoritarian legacies'. In the fourth section, we present the results and discuss their implications for the study of political legitimacy.

Socialization, generations and democratic legitimacy
The question of differences in political behavior across generations is one strand of an extensive literature on political socialization that originated in the 1950s. A pioneering and highly influential contribution to this literature was the work of Karl Mannheim, who argued that individual political attitudes and behavior are shaped by historical experiences in youth. Mannheim's framework had a strong impact on later studies, which set out to test his propositions empirically.
The earliest studies in the field of political socialization emphasized the importance of political learning in infancy, given the importance of the family as the primary agent of socialization (EASTON and DENNIS, 1969;HYMAN, 1959 importance of other socializing agents, such as universities, workplaces, or other specific political contexts, like elections (JENNINGS and NIEMI, 1968, 1974, 1981JENNINGS and MARKUS, 1984;SEARS and VALENTINO, 1997). Our study shares this broader view of political socialization, understood as a continuous process in which attitudes developed early in life are either reinforced or altered by later experiences.
Research on political culture and democratic legitimacy has incorporated theories about socialization and generations as a fundamental element in explaining changes in values over time and in relation to democracy. In this sense, the study by Inglehart and Welzel (2005) is particularly notable. Analyzing data from the World Value Survey (WVS), they found that younger cohorts, socialized in countries with higher levels of social and economic development, are more likely to develop values of 'self-expression', such as tolerance, and political participation in demonstrations, boycotts, and other non-traditional means of participation. In another study, Dalton (2004) shows that support for democracy is increasing in advanced industrial societies, especially among young people, despite their growing dissatisfaction and distrust of democratic institutions.
However, other studies point in the opposite direction. Foa and Mounk (2016), for example, identify a decline in support for democracy in the United States and Europe in the latest WVS data. More specifically, they find that older cohorts in those countries are more likely than younger cohorts to agree that it is very important to 'live in a democratic country'.
However, none of these studies includes 'new democracies'. Unlike the context of the old democracies, which often underwent slow and incremental processes of construction, more recent democracies were established just as the state itself was being rebuilt, and often in a context of economic, social and political upheaval. We can add to this the long experiences of authoritarianism and complex processes of political transition that these countries went through (DIAMOND, 1999;MOISÉS, 1995;TORCAL and MONTERO, 2006

Generations and support for democracy in Latin America
In studies on post-communist regimes in Eastern Europe, the question of generational effects on democratic legitimacy is central to the work of Mishler and Rose (1999. According to them, political legitimacy in the region is The thesis that older generations retain a sense of loyalty to the authoritarian regimes in which they were socialized makes sense when this experience refers to the collapse of an entire economic and social order that provided stability and material security to individuals, as is the case of the post-communist states studied by Mishler and Rose (2007). However, this context is far from universal. Latin America, for example, presents a much more heterogeneous scenario, with different types of military regime.
In this context, it is more difficult to devise a single model capable of explaining the effect of the political context on different generations.
Unfortunately, there are still few studies that identify generation as a key explanatory variable for understanding attitudes towards democracy. In general, researchers have at most included the variable 'age', though typically this has been more to control for its effect on other variables of interest than to explore its own explanatory potential. Booth and Seligson (2009, p. 123), for example, use age as a variable in their model on the determinants of support for democracy, finding no correlation between the two variables. Rennó et al. (2011), by contrast, found a significant relationship between age and support for democracy in Brazil, Venezuela and Ecuador. Salinas and Booth (2011) also found age to have a significant effect when testing the determinants of support for democracy in Latin America. None of these authors, however, analyzed the generational question or explored these results in greater depth.
Other studies have addressed the generational question, but not comparatively for the whole region. In his thesis, Del Porto (2012)  In a recent study, Moreno and Lagos (2016) gave centrality to the generational question. After more than three decades of the third wave of democratization, virtually all countries in the region are democracies. In this context, the authors ask themselves: does socialization under a democratic regime have an impact on citizens' attitudes?
To answer this question, the authors compared the democratic attitudes of two  (RUSTOW, 1970).
Less obvious, however, is the effect of authoritarian legacy. Following the same line of reasoning, we might expect it to have the inverse effect of the democratic legacy.
In this case, the greater the authoritarian legacy of a country, the less adherence we would expect to see to recently implanted democratic regimes.

Data and construction of variables
Our study uses the 2012 Americas Barometer, conducted by the Latin American Public Opinion Project (LAPOP) in 26 countries. In most countries, the sample consists of 1,500 interviews with respondents selected using probabilistic selection stratified across multiple stages. Canada, the United States, Belize, Guyana, Suriname, Trinidad and Tobago and Jamaica were excluded from the analysis, as they were not considered to belong to the region. Bolivia and Venezuela were also excluded, because for these countries data was not available for some of the variables we included in our model.
The first empirical challenge for the study was to construct a variable that would delineate different generations. We start from Mannheim's (1952) key proposition, that generations are formed based on the significant events that they have lived through. Using this definition, it should be possible to define different 'generations' by reference to the key events they have experienced. In our study, the event in question is a wide-ranging and drawn-out phenomenon: a transition in the political regime. As such, we compare two generational groups: one that socialized exclusively in a democratic political environment, and one that experienced two regimes, democratic and authoritarian.
It is true that many countries have had more than one political transition in recent history, and that processes of transition are varied and complex. However, we believe that our classification captures what we consider to be of greater importance in the current study, namely the experience of having, at some point over the life course, lived under an authoritarian regime.
From an operational point of view, the first step is to identify the 'cut-off point' that initiates the formative stage at which individuals are most susceptible to political learning. Since Mannheim (1952) there have been extensive debates on this question in the political socialization literature (JENNINGS and NIEMI, 1968, 1974, 1981 authoritarian, semi-democratic and democratic. The authoritarian regimes received the score zero (0), semi-democratic regimes a half point (0.5) and democratic regimes a point (01). The measure of a country's democratic legacy is therefore the sum of this score for the whole period 1945-2012. As such, the democratic legacy of a country will be greater when democracy has functioned for longer, and when it has worked better. The political legacy, however, is not restricted to the democratic legacy. Equally important for explaining individual attitudes, especially in relation to generational differences, is the authoritarian legacy. Most research on political culture has been restricted to considering how the duration and quality of democratic regimes affect support for democracy and democratic principles (BOOTH and SELIGSON, 2009;CHU et al., 2008). The authoritarian legacy, in turn, has received little attention. We believe, however, that it is relevant for precisely the same reasons: experiences of authoritarianism may create contexts that either favor or inhibit attitudes of support for democracy. These experiences also depend on their duration and nature, which, in the case of authoritarian experiences, are linked to the intensity with which civil and political rights are restricted.
In constructing our measure of the authoritarian legacy, we also started from Mainwaring's and Pérez-Liñán's (2013) classification, identifying all the years in which a country was classified as authoritarian, therefore receiving a score of zero (0). However, this classification does not distinguish between more and less repressive authoritarian regimes. For this reason, and so that we could more effectively assess the authoritarian legacy, we combine the latter with the measure of autocracy elaborated by the Polity IV project (MARSHALL and JAGGERS, 2007). Thus, for each year in which a country was classified as authoritarian by Mainwaring and Pérez-Liñán (2013), we replaced its zero score (0) with the Polity IV score, which ranged from 0 to -10. Under this measure, the closer the regime scored to -10, the more autocratic it was considered to be. Next, we In order to measure 'support for democracy', our dependent variable, we use a traditional question from the literature: "Now, moving on, with which of the following three sentences do you tend to agree with: For people like me, (1) There is no difference between a democratic or an undemocratic regime, or (2) Democracy is preferable to any other form of government, or (3) In some circumstances, an authoritarian government might be preferable to a democratic one". We recoded the responses to obtain a binary variable, where options 01 and 03 = 0 (does not support democracy), and option 02 = 01 (supports democracy). Table 02 shows the level of support for democracy among Latin Americans, presented as percentages. Columns one and two show the percentages for the generation that lived under the military regime and for the generation that has only lived under democracy, respectively, while column three shows the difference between the two generations. We also performed a chi-square test to verify whether these differences were statistically significant. The percentages of the fourth column marked (2018)  We note that in almost all countries, the group that experienced the authoritarian regime expresses greater support for democracy than that which was socialized exclusively under the democratic regime. In only three countries is this trend reversed, and of these exceptions only the case of Panama is statistically significant.

Results and discussion
Already, this result suggests that older generations that were socialized under the authoritarian regime are more likely to support the democratic regime. However, surely other variables also help to explain support for democracy. It is therefore necessary to construct a multivariate explanatory model.
The literature highlights the importance of socioeconomic variables in accounting for individual attitudes towards democracy. The most important of these is, undoubtedly, level of schooling, which is well established as a key determinant not only _____________________________________________________________________________________________________ 1 We recognize the inherent limitations of an observational research design of this kind. Ideally, experimental or panel studies would be drawn upon to help assess the causal relationships between the variables included in our model. Unfortunately, such data remain scarce for Latin America. Despite these limitations, we believe it is important that we seek to advance the field with the data and methods available to us. of support for democracy, but of political behavior in general (BOOTH and SELIGSON, 2009;NORRIS, 1999). In addition to schooling, the literature identifies another factor highlighted by modernization theory: place of residence, with people living in urban areas more likely to support democracy than those living in rural areas (MOISÉS, 1995).
Although it has received little attention in the broader literature, we also chose to add gender as a variable in our model as it has been identified as a factor explaining support for democracy in some Latin American studies (MOISÉS, 2008;SALINAS AND BOOTH, 2011). In addition to socioeconomic factors, other individual-level characteristics have also been highlighted. Interest in politics has, alongside education, been identified as one of the best predictors of an individual's political behavior, including democratic attitudes (DALTON, 2004(DALTON, , 2009). Finally, we add the individual's assessment of the country's economic situation. Studies have shown that adherence to democracy has 'instrumental' as well as 'intrinsic' motivations. That is to say, people tend to profess democratic values when they see improvements in their lives that they associate with a presiding democratic regime (BRATTOM and MATTES, 2001).
As the democratic legacy and the authoritarian legacy are contextual variables, we estimate multilevel logistic models for the 17 Latin American countries. Our level 01 and 02 model assumes the following form: Level 01: Y = B0 + B1*(lives in urban area) + B2*(male) + B3*(years of schooling) + B4*(interested in politics) + B5*(positive evaluation of country's economy) + B6*(generation) + R Level 2: B0 = G00 + G01*(authoritarian legacy) + G02 *(democratic legacy) + U0 In model 02, we include the interaction between generation and authoritarian legacy, and, in model 03, the interaction between generation and democratic legacy. Using this multilevel model, we were able to verify that the generational variable has positive and significant effect in both cases. As compared to individuals who were only socialized under democracy, individuals who had also lived under the military regime were shown to be 68% more likely to believe that democracy is always preferable to other forms of government. The persistence of such generational differences runs counter to the thesis that generational differences decline as democracy is consolidated (MISHLER and ROSE, 2007). At least in the context of Latin America, the decades of relative democratic stability experienced in most countries have not served to level out (2018)  Control variables generally performed as expected: 01. more years of schooling, interest in politics, and a positive evaluation of the country's economy were all associated with support for democracy; 02. being male had a positive effect, but it was not statistically significant; 03. although it is not the focus of our study, the finding that rural dwellers were more likely than their urban counterparts to support democratic regimes is surprising and deserves further investigation.
On the interaction between generation and authoritarian legacy, we can return to our hypothesis, based on the 'aversion argument'. We found no statistical significance in this relationship. We cannot, therefore, affirm that the generational effect on support for democracy is conditioned by the authoritarian legacy. In other words, we found no evidence to support the claim that where authoritarian regimes were more oppressive and longer-lasting there is a greater difference in support for democracy between older and younger generations. In fact, the authoritarian legacy has a negative effect on support for democracy 2 .
Each additional unit in the authoritarian legacy reduces support for democracy by 0.2%.
The democratic legacy, by contrast, has a positive effect: each additional unit increases likelihood of support by 0.2%. As the measure of democratic legacy varies from 0 to 65, and that of the authoritarian legacy from 0 to 363, a citizen of the country with the strongest democratic legacy is 13% more likely to support democracy than a citizen of the country with the weakest legacy, while a citizen of the country with the strongest authoritarian legacy is 72% less likely to support democracy than a citizen of the country with the weakest authoritarian legacy.
Our findings shed light on the generational question in Latin America from a perspective that is different, but complementary, to Moreno and Lagos' (2016). The comparison between the generation that has only experienced democracy and those that were also socialized under authoritarian regimes shows that the latter do indeed express a greater commitment to democracy.
This result fits with the contrast that Moreno and Lagos (2016) found between older and younger 'democratic cohorts'. What is curious, however, is that a greater authoritarian legacy does not increase the gap between generations. The crucial factor, then, is whether a generation has some experience of an authoritarian regime, regardless of how severe and repressive it was. We also found evidence that general context matters: a greater democratic legacy promotes support for democracy, while a greater authoritarian legacy discourages such support.
Taken together, the data reveal the explanatory power of political legacies.
Countries that combine democratic longevity and quality constitute an environment in which political socialization favors the formation of attitudes in support of the regime.

Conclusions
Most studies of democratic legitimacy ignore the generational dimension, merely using the 'age' variable as a demographic control in statistical models. By contrast, our study sought to emphasize the key role of generation in the formation of political attitudes, especially in new democracies. With this aim, we analyzed the generational effect on support for democracy in Latin America, comparing citizens who have only lived under democratic regimes with those who also experienced authoritarian regimes.
First, we find that,, it was possible to examine generational differences across the region and whether, to some extent, they were influenced by the countries' varied political legacies. As Moreno and Lagos (2016)  We expected that, in line with this result, the level of support among those who lived under authoritarian regimes would grow according to the intensity of the authoritarian legacy. However, empirical tests do not corroborate such an association.
By themselves, political legacies do have a broader effect on society, increasing, in the case of the democratic legacy, and diminishing, in the case of authoritarian legacy, the likelihood of support for democracy.
One important implication of these findings is that, in Latin America, generational replacement is not likely to lead to an increase in support for democratic regimes. In any case, the question of democratic legitimacy is more complex and deserves further investigation. This will allow a better understanding of generational differences in support for core democratic principles, such as tolerance, participation, constraints on power and the rule of law.
A second implication of our findings is that political legacies matter. Countries with weaker democratic legacies or stronger authoritarian ones will find it more difficult to generate a 'reserve of legitimacy' (DAHL, 1997). This is not to say that these e0001 -18/22 democracies are doomed to fail, but rather that, because of the long-term nature of the political legacy, the survival of democratic regimes are likely to depend more on institutional and economic factors.
One of the limits of the present study relates to the way authoritarian legacy has been measured. In order to move forward, we intend to elaborate more refined measures that take other dimensions into account, such as the level of human rights violations, number of people killed and disappeared, and the level of protest tolerated under the authoritarian regime. This will allow a more adequate comparison between different modalities of authoritarianism and their impacts on the political culture. In addition, our hypotheses should be tested with a larger and more heterogeneous sample of countries, and in other regional contexts.
Another important point is that our study considered only two generational groups: that which was fully socialized under the democratic regime, and that which lived under both democratic and authoritarian regimes. We believe that future studies can deepen analysis of both generations, taking into account the nuances that mark the political history of different countries in Latin America, many of which have undergone more than one authoritarian experience and process of democratization.
Similarly, different countries underwent different processes of political transition, some of these brief, but many of them drawn-out. As such, in addition to the experiences that individuals may have had with authoritarian and democratic regimes, there is a third context of socialization: in the transition process, itself. Furthermore, it is important to remember that in some countries recent events have raised questions regarding the democratic nature of their current political regimes, such as Peru, Honduras and Venezuela. It is thus also important to examine the possible implications of these experiences for democratic legitimacy.
Last but not least, future studies can also make analytical gains by using longitudinal data to see whether generational differences are decreasing or increasing over time.
Translated by Matthew Richmond Submitted on January 16, 2017 Accepted on November 06, 2017