1 A call of duty in hard times. Duty to vote and the Spanish Economic Crisis Carol Galais (carolgalais@gmail.com), Université de Montréal. André Blais (andre.blais@umontrea.ca), Université de Montréal. Abstract Does civic duty to vote fluctuate over time? Is it affected by changing individual economic conditions and/or by perceptions of the economy? The paper contributes to the literature on the effects of the global financial crisis on political attitudes by focusing on duty to vote, a central element of the democratic culture. We examine, through a four-wave panel survey, the effects of the Spanish economic crisis on feelings of civic duty. We empirically test the assumption of early formation and resistance to change, and then determine whether or not an economic crisis can, in addition to weakening support for authorities and institutions, erode citizens’ sense of civic duty. We find that civic duty to vote did not decrease during the economic crisis and that it was quite stable at the individual level, though more so for middle-aged adults than for younger citizens. For individuals under 30 years of age we find a positive association between changes in income and changes in duty, that points to some negative effects of the crisis on their levels of civic duty. Contrarily, middle-aged adults were not affected at all by changes in their personal work situation or evaluations of the economic situation. 1. Introduction Civic duty is the missing link of the rational choice equation for turnout, the attitudinal element that is able to predict if people will show up at polls even if the costs of voting exceed the benefits (Blais 2000). Besides being a predictor of turnout of utmost relevance, it is also an attitude related to citizenship norms, political system legitimacy, support and engagement. It is therefore believed to be deeply rooted in early socialization and not to change much over the life span. Yet, a growing concern about the effects of the Great Recession points to possible deep changes in the political culture (Vaiman et al. 2010, Stevenson and Wolfers 2011, Chippendale 2012). In many western countries the Great Recession has triggered a debt crisis followed by austerity measures in which citizens had no say (Armingeon and Baccaro 2012). As a consequence, several political attitudes, such as trust in democratic institutions, are believed to have endured serious damage (Armingeon & Guthmann 2012). These effects could run deeper in new or consolidating democracies, where support for representative democracy or trust in elections could be threatened (Córdova & Seligson 2009). It is therefore worthy to look at the possible effect of the economic crisis on an attitude that is directly related to the decision to participate or not in elections and indirectly to the very legitimacy of the political system. When people view voting not only as a right but also as a moral obligation towards the community, they implicitly acknowledge that they are a member of that society and that there are duties as well as rights associated with such membership. The question is whether people are less inclined to think in those terms when the economic situation takes a sharp downturn. 2 In order to explore the effects of the economic crisis on this attitude we are using data coming from Spain. This is one of the European countries hardest hit by the debt crisis. The economic situation that the country is suffering since 2008 has affected the lifestyles of many families, and probably also the way in which citizens perceive the state and politicians. It is still to be seen if the economic crisis has also undermined the psychological basis of political engagement and some of the foundations of the political system legitimacy such as the civic duty to vote. For these purposes, we are using panel survey data that allow us to track changes in Spaniards’ duty to vote over a period of a year and a half (November 2010-May 2012) when major cutbacks and austerity measures were implemented. The present paper addresses four questions. First, does civic duty to vote vary over time at the aggregate and/ or at the individual level? Second, do changes in one’s economic situation affect civic duty to vote? Third, do individuals’ perceptions of the economy have an effect on this attitude? Fourth, and last, is the evolution of this attitude related to other factors such as the individual’s age or the political context? To address these questions, we first review the literature on the stability of attitudes belonging to the attitudinal dimensions of legitimacy and political support. We also examine previous work on the role of contextual factors, particularly the economy- in the evolution of such orientations, as well as previous work on the relationship between age and contextual effects. Next, we present the research design and the data, along with the models used to estimate change in the duty to vote. We then present the empirical evidence. After some descriptive analyses we run a fixed-effects panel model to estimate the impact of objective and subjective economic conditions on individuals’ feelings of civic duty during the economic crisis. 2. Attitude change, civic duty and the economy Studies in political behavior and political culture have traditionally observed attitudinal change with suspicion since Converse (1964) raised the possibility that survey answers could be random, as not every individual develops opinions on every political object. Under this perspective, changing attitudes could indicate uncertainty or even the absence of attitudes. But, as many methodologists have demonstrated, attitude instability may also be due to poor measurement instruments that introduce measurement error (Achen 1975, Erikson 1979; Alwin & Krosnick 1991; Heath, Evans and Martin 1994). Does this means that genuine attitude change does not exist? Not at all. There has been overwhelming and stubborn evidence of attitude change, and more specifically of the influence of context on political orientations. Among those contextual factors able to trigger attitudinal change one clearly stands out: the economy. Indeed, the link between the political culture and the economy has been a fruitful subject in Comparative Politics. Modernization theory, for instance, posits that prosperous economic conditions trigger improvements in socio-economic conditions, which in turn pave the way for a cultural change consistent with Democracy (Lipset 1959, Deutsch 1961, Burkhart and Lewis-Beck 1994). Building on this idea, Inglehart formulated his theory of cultural change (1997). According to Inglehart, western societies evolved from traditional to modern culture as a consequence of the industrial revolution, and from modern to post-modern during the 60s as a consequence of sustained well-being, as growing up in prosperity stimulates the adoption of post-materialist values. Such reasoning could apply as well to the civic duty to vote: when basic needs are fulfilled, it is possible for people to pay attention to the social norms that are beneficial for the larger community. 3 However, the influence of the economy on political culture is not in a single direction, since economic success is not irreversible. Indeed, the economy can deteriorate and this can foster authoritarianism (Duckitt and Fisher 2003) or intolerance towards immigrants (Kessler y Freeman 2005, Rink et al. 2009, Armingeon and Baccaro 2012). The role of economic downturns has been recently revisited in the light of the 2007 Great Recession (Vaiman et al. 2010, Chippendale 2012). Stevenson and Wolfers (2011), for instance, point out that this particular crisis has undermined trust in public institutions in different countries. Armingeon and Guthmann (2012) show a relationship between negative perceptions of the economic crisis and a decrease of trust in parliaments. These detrimental effects could be larger in consolidating democracies, going as far as damaging citizens’ trust in elections or support for representative democracy (Córdova & Seligson 2009). The question to be examined is whether the economic crisis has similar negative effects on duty to vote as on those aforementioned attitudinal dimensions. In this respect, two theories may help filling the gap between macro-economic factors and individual attitudes related to democracy: the socio- cultural and the institutional performance models (Newton & Norris 2000). The first emphasizes the role of individual life situations and experiences in shaping civic mindedness and other attitudes consistent with democracy. When referring to the effects of economy, personal experience means mostly labor market transitions (i.e. entering the labor market, becoming unemployed, a change in the labor status…). These situations promote attitudinal change by altering individuals’ social networks and economic interests, defining their new role and even triggering the transition to adulthood and, thus, the internalization of social norms (Marini 1987, Scott 1995, Shanahan 2000). With regard to civic duty, a worsening in individual’s economic and working conditions may induce citizens to focus on their own personal interests rather than those of the community, weakening the social norm that voting is a duty. An economic crisis may also alter the normal sequence of events in the life cycle - first job, improved purchasing power, having children, owning a home…- delaying or hampering the transition to adulthood and the new interests in the polity that come with it. On the other hand, the institutional performance model suggests that perceptions of economic and political conditions are the link between macro factors and citizens’ orientations (Almond and Verba 1963, Weatherford 1992), connecting institutional performance and political legitimacy. For instance, Štulhofer (2002) brings forth that social changes stemming from an economic crisis affect evaluations of economic, political and social conditions, which in turn affect dominant values and foster cynicism and opportunism. From this perspective, a worsening of the economy would adversely affect duty to vote as a consequence of disapointment with the performance of the system. Thus, our general expectation is that a worsening of the economy could erode individuals’ civic duty to vote. Yet we think it is necessary to control for two elements when it comes to understanding how the economy may bring forth attitudinal change: the political context and the susceptibility of attitude change due to age. Concerning the political context, it makes sense to think that some citizens will “refresh” or update their sense of duty as an election approaches, as they do with respect to confidence in democracy and political institutions, at the time of elections (Franklin 2004, Bowler & Donovan 2002). Actually, some attitudes refering to the system as a whole have been found to react more to to political factors such as electoral processes than to economic conditions (Evans and Whitefield 1995, Bratton and Mattes 2001). It may well be, for instance, that the presence of an election will more than offset the negative effect of the economic crisis. 4 Another element that must be taken into account is age susceptibility to attitudinal change. Indeed, the socializing effects of context, including those due to elections or the economic crisis, may not be homogenous for all ages. Franklin, for instance, argues that the electoral context has a greater effect on younger citizens (2004). Between the "lifelong openness" hypothesis -which considers attitudinal change to occur across the whole life cycle- and the "persistence" hypothesis -according to which most fundamental political attitudes are early formed and reluctant to change-, there are hypotheses that condition the likelihood of attitudinal change to age to a greater or lesser extent. The first perspective is the "increasing persistence" theory, which considers attitudinal change quite likely In early adulthood, followed by gradual resistance to change as the individual gets old (Glenn 1974, Berelson, Lazarsfeld, & McPhee 1954; Newcomb, Koenig, Hacks, & Warwick 1967). The “impressionable years” hypothesis suggests that the individual is open to external influence as long as this takes place between adolescence and early adulthood, when the role of the individual in society begins to be defined. The attitudes originated during this period would be pretty stable for the rest of the individual's life (Mannheim, 1952; Sears 1975, Markus 1986, Alwin and Krosnick 1991). Finally, the theory of symbolic politics makes attitudinal change across life dependent on the nature of the attitudes. Those orientations developed on an affective basis dealing with diffuse political symbols would stand out for their early formation and stability. In other words, symbolic attitudes would behave according with the “persistence” theory, displaying great stability over the life cycle (Glenn 1980, Sears and Funk 1999). The idea that there is a group of “stable” attitudes –symbolic, normative, affective and early formed- and a group of “volatile” attitudes -fundamentally evaluative and likely to be sensitive to context-, is widely accepted (Klapper 1960, Sears 1975, Edwards 1990, Crano & Alvaro 1997, Fabrigar & Petty 1999, Ajzen 2001, Goren 2005). Civic duty to vote would belong to the first group, as it has been described as a moral attitude towards the democratic regime, analogous to the Political Regime’s legitimacy (Fuch 1993). Similarly, Weatherford considers the civic duty to vote an element of political involvement, which belongs to the citizenship traits of political legitimacy (1992:154). Civic duty can also be understood as the belief that the good citizen should vote in a democracy and that it is thus a moral obligation to participate at least minimally in the electoral process (Blais 2000, Blais and Achen 2012). Therefore, the belief that voting is a duty is transmitted trough primary socialization along with a set of basic attitudes and values (Schwartz 1994, 2005). As a normative attitude, it is shaped by socialization agencies that operate in early stages of life, such as the family or the school (Jennings, Stoker and Bowers 2009, van Deth et al 2011). According to these hypotheses, the economic crisis would have not played any role at all in the evolution of the civic duty to vote among Spaniards (persistence hypothesis, symbolic politics), may trigger some change across the whole population (lifelong openness), more change among youngsters than among the older (increasing persistence) or change only among those who happen to be in their impressionable years. As a crisis of this magnitude is likely to have altered people’s expectations with regard the labor market as well as how they see themselves in society and the responsibilities they have as citizens, we would argue that, if bad economic conditions or perceptions undermine civic duty, the impact should be felt first and foremost among those in their impressionable years. It should be at this time, when individuals become entitled to new rights and responsibilities, that they start paying more attention to public issues, and that perceptions of the economic and political context shape present and future attitudes (Boninger, Krosnick, & Berent 1995). 5 3. The Spanish economic recession and duty to vote. In order to ascertain the effects of the economic crisis we need longitudinal data that tap the deterioration of macroeconomic factors as well as the electoral climate. For these reasons, we selected Spain as a case study. Spain is a relatively young democracy whose system legitimacy may still be under construction, therefore sensitive to major economic changes. Spain, as other Southern European countries, has been severly hit by the Great Recession, still suffering its consequences. A quick outline of the Spanish economic crisis takes us back to 2008, when the explosion of the real estate bubble made housing prices collapse. This suddenly stopped the housing construction, sending thousands of people working directly or indirectly in the construction industry to unemployment. The public sector has not done any better. 185,000 public jobs have been eliminated between January 2010 and July 2012 by means of retirements that have not been replaced. 1 Most of the cuts have been made in local administrations, and only in 2010 Spanish city councils reduced the number of their public employees by between 1% and 5%. As a result, unemployment rates jumped from 8 % in 2007 to 26% in 2012, 50% among young people. 2 Soon the banks found out that (subprime) mortgages could not been paid anymore and drastically cut on loans to business, entrepreneurs and individuals, which worsened the situation. Although the European Central Bank approved a bailout for the Spanish banks in 2012, credit had not yet begun to flow to individuals or enterprises. Furthermore the bailout came in exchange of the commitment to trimming the budget deficit to a 3.6% of GDP, which entailed further welfare cutbacks. To this turmoil in the Spanish economic situation we must add that a general election was held in November 2011, which resulted in government change. We face here contradictory forces regarding the expected effects of political and economic conditions on duty to vote. On the one hand, we could expect an increase in duty at the time of the election. On the other hand, negative perceptions of the economy and of politicians’ capacity to deal with it should operate in the opposite sense, hampering the belief that voting is a civic duty. 4. Research design and data. Since our research hypotheses entail longitudinal analyses we use a panel survey. This is study CIS 2855, an online panel survey conducted in four waves over a period of 18 months in November 2010, May 2011, November 2011 and May 2012. 3 Respondents were selected from an online survey pool set up through active recruitment of potential subjects in the main commercial on-line services and websites in Spain.4 This introduces deviations from a representative sample of the general population, although quotas were applied in the sampling process in order to reduce biases related to the non-probability nature of the sample. Last, but not least, the survey presents the habitual attrition rates, with 1,300 individuals out of 2,100 wave 1 respondents completing the wave 1 The public sector in Spain has been described as medium-sized, with 5.5% of their populatin employed in the public sector in 2008, which is average for the UE-27, similar to Italy (5.7%) or Germany (5.4%).1 2 The annual figures are: 11% in 2008, 18% in 2009, 20% in 2010, 22% in 2011. 3 The survey was sponsored and funded by the Centro de Investigaciones Sociológicas (CIS) and the Universitat Autónoma de Barcelona (UAB) research group “Democracy, Elections and Citizenship” (P.I. Eva Anduiza). 4 Further details can be found at http://www.netquest.com/papers/esomar26_en.pdf 6 4 survey.5 Because of the purposes of our study (tracking and explaining attitudinal change), we have not excluded any of these participants. This decision was made in order to minimize panel- conditioning effects, a phenomenon that happens when panellists change their attitudes or behaviours as a result of being on a panel study (Sobol 1959, Clinton 2001).6 The sample consists of Spanish citizens between 16 and 45 years of age with Internet access, as the survey sample was designed to tap attitudinal change among people who are relatively young and thus in theory open to change.7 In this respect, we can consider that the sample gathers two age groups: the younger, under 30 years old at the beginning of our study, and the middle-aged adults, between 31 and 45 years old when our study ended. Systematic comparisons between these two groups will allow us to test if attitudinal change is more likely to occur before one is in his/her thirties. The convention is to consider that impressionable years end at age 25 (Krosnick and Alwin 1989, Alwin et al. 1991), but lately the frontiers of what is considered “youth” have expanded so that 30 is now often viewed as the turning point, especially in the case of Spain (Garcia and Martin 2010), where in 2010 almost 60 percent of people between 16 and 34 years of age still lived in their parental homes (Aparicio and Oppedisano 2012).8 For the goal of tracking changes in the belief that voting is a duty we first examine the aggregate levels of this attitude, comparing the two different age groups, construed according to their susceptibility to attitude change. Next, we move to examining intra-individual change, paying attention to individuals’ transitions between the waves of our panel. For this purpose, transition probability matrices are used to report the chance that an individual changes her mind regarding voting being a duty; plus we tabulate the frequency of trajectories for this question across waves. Last, we examine the effects of the aforementioned factors of attitudinal change. To test our longitudinal hypotheses we estimate over time change in individuals’ duty to vote by means of a series of fixed-effects OLS panel estimations that we replicate for both age groups. Fixed-effects models explore the relationship between a dependent variable and a set of factors that change over time. The fixed-effects models neutralize the impact of all time-invariant individual characteristics. This means that the resulting coefficients cannot be biased because of omitted time-invariant characteristics such as gender, race, religion, previous socialization and age or education. The equation for the fixed effects model is: Yit = β1 Xit +αi + uit Where :  αi (i=1…n) is the unknown intercept for each individual, and there are n-individual- specific intercepts 5 An additional 620 freshly recruited respondents entered the study in the second wave to compensate the fact that our sample was overeducated. All in all, 2720 individuals were part of the study at some point in time. 6 We have tested our models on a sample consisting on these 1300 persons and found higher rates of change. Thus, by keeping as many respondents on our sample, we are minimizing those panel conditioning effects, 7 Respondents were aged between 16 and 44 in wave 1 and 17 and 45 in wave 4. 8 We could have compared those aged between 16 and 25 against the rest of the simple and none of the forthcoming analyses would have yield dramatically different results. Yet the number of cases will reduce a lot in some cases. For instance, in wave 4 we can count 574 individuals that had less than 30 years old at the beginning of our study, but only 268 that had less than 25. 7  Yit is the dependent variable where i= individual and t=time  Xit stands for an independent variable  β1 is the coefficient for that independent variable  uit is the error term. Thus, for a given individual, as X varies across time by one unit, Y increases by β units (Bartels 2008). Among the advantages of this estimation method we can list the large amount of observations, which increases degrees of freedom, reduces collinearity among the explanatory variables and improves the efficiency of the estimates. Turning to the question of measurement, we have used the Blais (2000) and Blais & Achen (2012) operationalization for duty to vote. This is a 4 category ordinal variable resulting from a set of two questions. The first asks people whether they think that voting is duty or a choice. Those selecting the “choice” option are coded as 0. Those selecting the “duty” option are next asked to what extent they believe that voting was a duty: a little (1), some (2) or a lot (3). The question is designed to avoid social desirability effects that would result in an overestimation of civic duty. It offers a “positive” alternative (“voting is a choice’) for those who do not feel a duty to show up at the polling station.9 All independent variables in our models have been recoded so that they range from 0 to 1. Thus, their coefficients can be interpreted in terms of the effect on duty across time when they move from its minimum to its maximum value. We consider three basic groups of variables. First, the economic situation and working status of the respondents. Second, their perceptions of the economy, including evaluations of politicians’ capacity to deal with the economic crisis. Finally, we include controls for political context, operationalized as dummies for each panel wave that should be interpreted as the average increase in the levels of the dependent variable, compared to wave 1, everything else being equal. We also account for susceptibility to attitude change due to age replicating the analyses for the “young” (under 30 years old at the beginning of our study) and the middle-age adults (between 30 and 45 years old in our study). 9 The exact wording is : A. For some people vote is mostly a duty. They think they should vote regardless of their views on the parties. For others, voting is an option. They decide whether or not to vote according to their opinions about the parties. For you voting is all about: 1 A duty (→ ask B) 2 An option B. To what extent do you feel it a duty? 1 A lot 2 Some 3 A little 8 The deterioration of personal economic circumstances may have manifested itself in a number of ways at the individual level. In the first place, we consider income, a 11-category variable ranking from 0€ to more than 6.000€ per month. Next, we take into account two conditions that imply a change related to occupational status: becoming employed/ unemployed and gaining/loosing the status of permanent worker. While acquiring a job fosters social integration, provides new social networks and puts people in touch with dominant values becoming unemployed leads to social exclusion and frustration with the system. We also consider changes in the “permanent” worker condition, with the expectation than becoming temporary worker entails a higher degree of uncertainty in making ends meet. The next set of variables is related to perceptions of economic conditions. The first question asks whether the economic situation is perceived as very bad, bad, fair, good or very good. Next, the questionnaire asks respondents’ perception on how these general economic conditions have evolved during the last year. The following question is about the respondents’ forecast about the economic situation for the upcoming year. The last indicator asks about the individual’s evaluation of her personal economic conditions. The possible answers for these three questions are worst, the same, and better. The survey also asks about the perceived competence of the two main parties (PSOE, the socialist party, ruling until the third wave of the panel, and PP, the conservative party leading the opposition until November 2011). The possible responses are: very bad, bad, fair, good, or very good. We have recoded all these subjective evaluations so as they all range from 0 to 1. In the next section, we explore the amount of change observed over the course of the survey (a year and a half). We then move to the multivariate, longitudinal analyses of individuals’ feelings of civic duty. 5. Results Figure 1 compares the evolution, between November 2010 and May 2012, of duty to vote for the young and the middle-age adults. The graphs depict the means at each point in time –taking into account that this variable ranks between 0 and 3-, as well as the 95% confidence intervals. We see slightly different patterns which do not seem consistent with the hypothesis that the economic crisis hampers civic duty. The younger group follows a steady slightly increasing trend, especially between waves 2 and 3, coinciding with a pre-electoral context. They start with a lower average of civic duty to vote and end with a (slightly) higher average than the middle-aged adults. The latter follow a more stable pattern, with a slight decrease in their average levels of duty in wave 2. Bonferroni tests reveal non-significant differences in means for all the possible pairs of waves for the middle age group age. On the contrary, the younger group age exhibit significant differences between waves 1 and 4 among (at 95%), and between waves 1 and 3 (at 90%). (FIGURE 1 ABOUT HERE) From an individual perspective, the panel data allow to examine the transitions between categories of a variable across waves. Tables 1 to 4 offer a first perspective on the amount of stability/change over the course of the study. We should read the results for each cell in Table 1 in terms of the chances that an individual answering “None”, “Little”, “Some” or “A lot” will change her mind in the next panel wave, taking into account that a reported probability aggregates transitions from 9 wave 1 to wave 2, from wave 2 to wave 3 and from wave 3 to wave 4. 10 Thus, the belief that voting is an option has a 74% of chances of remaining stable across waves among the youngest, and 77% among the 31-45 age group. The most dutiful option is almost as stable for both age groups (69-71% respectively). Yet, those with low levels of duty are likely to change either to “none” or “some” answers, the pattern being similar for both young and middle-aged adults. “Little duty” is the category most likely to change, either to “none” or to “some” duty. Intermediate levels of duty (“some”) can also easily turn into high levels of duty (this happens in 29% of the cases among the younger, 27% among the middle-aged) or into no duty at all (24% of the times for both age groups). (TABLE 1 ABOUT HERE) (TABLE 2 ABOUT HERE) Tables 3 and 4 tell us that 46% of the observations -both younger and older- are classified as “not dutiful”, this being the most frequent case in our database. The ‘between’ columns tell us how many citizens were found in a given category at least once. Thus, 62% of the younger selected the “option” answer at least once, versus 60% of the middle-aged. This is a small difference, and the same pattern is found for the rest of the categories. The total percentage in this column shows that 56% of younger individuals experienced some degree of change, selecting different categories on the duty question in different waves. 48% of the middle-aged changed their mind over time, indicating slightly more stability in that group. The ‘within percent’ computes the fraction of time that a respondent has a specific value of duty. Thus, if an individual ever said that voting is a choice, 75% of her answers will be « choice » if she is younger than 30, and 78% if she is older than 30. If a young individual ever said that she feels strongly the duty to vote, then 66% of her responses would point in that same direction, a percentage that slightly increases to 68% if she is older than 30. Hence, the civic duty to vote is overall pretty stable, though somewhat slightly more stable among middle-aged adults. 11 This is in line with the impressionable years hypothesis. We should keep in mind that all these differences are quite small, although they suggest that contextual factors -such as the economy or elections- may have some impact on the evolution of civic duty, particularly among younger citizens. (TABLE 3 ABOUT HERE) (TABLE 4 ABOUT HERE) We now proceed to isolating the independent effects of the economic crisis through individuals’ changing conditions. Coefficients from four different specifications are reported in Table 5. The analyses have been replicated for both age groups. The first specification (models 1 and 3) only accounts for objective changes in the economic and work status of the respondents that occurred during the study. The second (models 2 and 4) adds perceptions of the economy, including perceptions of the capacity of the two main parties to deal with the crisis. 10 The « xttrans » stata command produces the probability that xi,t+1 = ν2 given that xit = ν1 by counting transitions. 11 We have conducted the same analyses for interest in politics and the intensity of party identification and found that responses are slightly more stable in the case of interest in politics, and slightly less for party identification. We can therefore conclude that civic duty is relatively stable, though not quite as much as interest in politics. 10 (TABLE 5 ABOUT HERE) All these models take into account the effects of context, as indicated by the presence of dummies for each wave, which is consistent with two-way fixed-effect panel model specifications. This is intended to account for unobserved heterogeneity across both time and individuals. With regard to period effects, all waves are significantly different from wave 1 . Waves 3 and 4 are also significantly different from wave 2, confirming the existence of the trends described when discussing figure 1. It is noticeable that the coefficients of waves 3 and 4 for the younger sample are almost twice the size for the older, suggesting a greater effect of the context for those that had less than 30 years old at the beginning of the study. From these coefficients we can infer a period effect –especially among the youngest- caused by the election held in November 2011, which took place one week after wave 3 was conducted. Interestingly, this effect had not decayed 6 months after the election. The magnitude of the period effects is about the same that can be observed in figure 1 which means that the slight increase in duty that took place during the course of the study cannot be imputed to changes in individuals’ economic conditions or perceptions. The first model includes the variables that capture the objective changes in the individuals’ economic and labor status. Since all the independent variables in the model are coded from 0 to 1, the coefficients indicate by how much duty increases when the variable shifts from its minimum to its maximum value. What we see is that changes in employment status do not seem to have any impact on whether one feels that it is her duty to vote in elections, for any of the age groups. This finding is consistent with what we have observed at the aggregate level, with overall levels of civic duty actually increasing during the economic crisis. Yet the story is different with regard to income. The significant positive coefficient for income in model 1 indicates that changes in income are followed by changes in civic duty in the same direction. In other words: younger citizens tend to experience a decrease of their civic duty to vote as their income decreases. On the contrary, the evolution of civic duty for the middle-aged is not significantly related to changes in income. Along with the coefficients related to changes in employment status, this indicates that the overall economic downturn was not accompanied by a general weakening of civic duty for the middle aged. The younger were more prone to question the norm that it is a moral obligation to vote if and when their economic status worsened. Among the 16-30 years old, 70 percent witnessed some change in income during the study, being negative changes more important in latter waves.12 When there was some negative change, the most frequent change was one of -0.1 on the 0 to 1 scale. According to our estimations, this would produce a decline of 0.035 on the (0 to 3) duty scale, a rather small effect. The bottom line is that most people’s income did not move substantially during these 18 months and the mean income scale declined only marginally, from 0.35 in wave 1 to 0.3 in wave 4 in the case of the youngest.13 More precisely, 22 percent of the younger experienced a decrease in their income between waves 1 and 2, whereas 21 percent experience an increase. These percentages move to 23 percent decrease, 20 percent increase between waves 2 and 3. Finally, between waves 3 and 4 the percentage of those who get worse is 24, while 17 percent experience an increase. Our results suggest that this decline ‘caused’ an overall decline of 0.2 on duty. This decline was more than offset by ‘other’ factors, the 12 61 percent of the individuals within the older age group experienced changes in income 13 The average income –measured from 0 to 1- in our simple decreases from 0.4 (wave 1) to 0.36 (wave 2), indicating that the general trend over time is downwards. The trend for the older is the same than for the younger (-0.05 points on average) but starting higher, at 0.45. 11 most visible being the national election, whose impact seems to have been larger than that of the economic crisis. The next block (models 2 and 4) includes economic perceptions. None of these variables reaches the level of significance, confirming that subjective perceptions of the Great Recession had no impact on feelings of duty. The same applies to the perceptions of the capacity of the two main parties to deal with the crisis: many respondents thought that the two main parties were incompetent but such judgments did not lead them to revisit their views about whether they had a moral obligation to vote in elections. 6. Conclusions In this paper we have addressed the question of the stability of civic duty to vote during an economic crisis, focusing on the Spanish case. We approached this issue by means of panel data and we presented empirical evidence about the evolution of this variable during hard times. Both modernization theory and recent works on the cultural effects of the Great Recession suggest that economic conditions –and particularly an economic crisis- may erode levels of political support and political legitimacy, especially in new democracies and among youngsters. We wanted to ascertain whether the economic crisis had a negative impact on Spaniards’ levels of duty to vote. Aggregate data indicate that this variable is pretty stable, although not completely static. Individual descriptive data on transitions point out that most of the change happens for people with intermediate levels of civic duty and that young adults are slightly more prone to change than middle-aged adults. Four fixed effects models tested the effects of objective and subjective economic conditions on the evolution of civic duty. We found no evidence that a worsening in one’s perceptions of the economy weakens sense of civic duty. The same applies to personal economic or working conditions for middle-aged adults, whose civic duty is resilient to the effects of the economic crisis. For the youngest, however, we detected a significant relationship between income loss and the deterioration of their belief that voting is a duty. It may be that a worsening in one’s economic and social status endangers the construction of the individual’s identity as a citizen, or that impoverished young citizens do not see the point in voting and participating in a system that is being tough on them. It may as well be that this entails some degree of “arrested development”, this is, a delay in the acquisition of norms and responsibilities that come with adulthood. In any case, we cannot discard an effect of the crisis in the formation and consolidation of this social norm. The presence of two different patterns for the two age groups with regard the civic duty to vote– resistance to change for middle-aged adults and some sensitiveness to economic conditions for younger citizens- is in line with the impressionable year hypothesis. This suggests that duty may behave as recent studies have pointed out for system legitimacy or support, but only among the youngest segment of the population. The symbolic nature of this attitude may foster its crystallization before adulthood, but not as soon as the « symbolic politics » theory predicts. That being said, the most important finding is that even among the 16-30 age group sense of civic duty did not decline during the economic crisis. There was a decline among those hardest hit by the crisis but that impact was modest, and completely offset by other factors, the most obvious being the presence of an election. This is good news in the sense that adhesion to civic virtues survives even in very hard times and in a country where democracy is still in a process of consolidation. Yet it 12 remains to be seen whether, over the long haul, a severe economic crisis could weaken citizens’ feeling that it is their duty to vote in elections. References Aguilar, E. E., & Pacek, A. C. (2000). Macroeconomic conditions, voter turnout, and the working- class/economically disadvantaged party vote in developing countries. Comparative Political Studies, 33(8), 995-1017. Ajzen, I. (2001). Nature and operation of attitudes. Annual review of psychology, 52(1), 27-58. Almond, Gabriel A. 1980. "The Intellectual History of the Civic Culture Concept." In The Civic Culture Revisited, ed. Gabriel A. Almond and Sidney Verba. Boston: Little, Brown. Alwin, D. F., & Krosnick, J. A. (1991). “Aging, cohorts, and the stability of sociopolitical orientations over the life span”. American Journal of Sociology, 169-195. Alwin, D. F., Cohen, R. L., & Newcomb, T. M. (1991). Political attitudes over the life span: The Bennington women after fifty years. Univ of Wisconsin Press. Alwin, DF and J.A. Krosnick (1991) “Aging, cohorts, and the stability of sociopolitical orientations over the life span” The American Journal of Sociology Aparicio, A., & Oppedisano, V. (2012). Fostering the emancipation of young people: Evidence from a Spanish rental subsidy. (No. 6651). Institute for the Study of Labor (IZA). Armingeon, K., & Guthmann, K. (2012). Democracy in Crisis? The declining support for national democracy in European countries 2006-2011. Armingeon, Klaus, and Lucio Baccaro. (2012). "The Sorrows of Young Euro: Policy Responses to the Sovereign Debt Crisis." In Coping with Crisis: Government Reactions to the Great Recession, eds. Nancy Bermeo and Jonas Pontusson. New York: Russel Sage Bartels, Brandom (2008) “Beyond “Fixed Versus Random Effects”: A framework for improving substantive and statistical analysis of panel, time-series cross-sectional, and multilevel data”, Stony Brook Uiversity, working paper. Berelson, B., Lazarsfeld, P. F., McPhee, W. N., Lazarsfeld, P. F., & Lazarsfeld, P. F. (1954). Voting. Chicago : Univ. Chicago Press. Blais, A. (2000). To vote or not to vote?: The merits and limits of rational choice theory. University of Pittsburgh Pre. Blais, André, Achen, Christopher H. (2012). “Taking civic duty seriously: political theory and voter turnout.” Princeton, mimeo. 13 Boninger, D. S., Krosnick, J. A., & Berent, M. K. (1995). « Origins of attitude importance: Self- interest, social identification, and value relevance ». Journal of personality and social psychology, 68, 61-61. Boulianne, S. (2009). Does Internet use affect engagement? A meta-analysis of research. Political Communication, 26(2), 193-211. Bowler, S., & Donovan, T. (2002). Democracy, institutions and attitudes about citizen influence on government. British Journal of Political Science, 32(2), 371-390. Brewer, Gene A. (2008). Employee and organizational performance. In Motivation in public management:The call of public service, ed. James L. Perry and Annie Hondeghem, 136–56. Oxford,UK: Oxford University Press. Burkhart, Ross E., and Michael S. Lewis-Beck. (1994). "Comparative Democracy: The Economic Development Thesis." American Political Science Review 88:903-10. Chippendale, E. (2012). The global financial crisis and public sentiment towardsiImmigration and immigrants in the Netherlands: implications for liberal democracy and political culture. Clinton, J. D. (2001, May). “Panel bias from attrition and conditioning: A case study of the Knowledge Networks panel”. In AAPOR 55th Annual Conference. Colombo, C., Galais, C., & Gallego, A. (2012). “El uso de Internet y las actitudes políticas: Datos cuantitativos y cualitativos de España. “Arbor, 188(756), 751-766. Converse, Phillip E. (1964). “The Nature of Belief Systems in Mass Publics.” In Ideology and Discontent, ed. David P. Apter. New York: Free Press. Córdova, Abby and Mitchell A. Seligson (2009). Economic Crisis and Democracy in Latin America. PS: Political Science & Politics, 42, pp 673-678. Crano, W. D., & Alvaro, E. M. (1997). The context/comparison model of social influence: Mechanisms, structure, and linkages that underlie indirect attitude change. European review of social psychology, 8(1), 175-202. Demmke, C. (2005). Are civil servants different because they are civil servants?. Présidence luxembourgeoise du Conseil de l'Union européenne: European Institute of Public Administration. Deutsch, Karl W. (1961). "Social Mobilization and Political Development." American Political Science Review 60:493-514. Diamond, Larry, ed. 1993. Political Culture and Democracy in Developing Countries. Boulder: Rienner. Duckitt, J., & Fisher, K. (2003). The impact of social threat on worldview and ideological attitudes. Political Psychology, 24(1), 199-222. 14 Edwards, K. (1990). The interplay of affect and cognition in attitude formation and change. Journal of Personality and Social Psychology, 59(2), 202. Erikson, R. S. (1979). The SRC panel data and mass political attitudes. British Journal of Political Science, 9(1), 89-114. Fabrigar, L. R., & Petty, R. E. (1999). “The role of the affective and cognitive bases of attitudes in susceptibility to affectively and cognitively based persuasion.” Personality and Social Psychology Bulletin, 25(3), 363-381. Franklin, C. H. (1984). “Issue preferences, socialization, and the evolution of party identification.” American Journal of Political Science, 459-478. Franklin, M. N. (2004). Voter turnout and the dynamics of electoral competition in established democracies since 1945. Cambridge University Press. Freeze, M., & Montgomery, J. (2012). Static Stability and Evolving Constraint: Attitude Stability, Constraint, and Impact in the American Electorate. Geoffrey Evans and Stephen Whitefield (1995). “The Politics and Economics of Democratic Commitment: Support for Democracy in Transition Societies”. British Journal of Political Science, 25, pp 485-514. doi:10.1017/S0007123400007328. Gilley, B. (2006). “The determinants of state legitimacy: Results for 72 countries”. International political science review, 27(1), 47-71. (Economic performance affects state legitimacy) Glenn, N. D. (1974). « Aging and conservatism. » The ANNALS of the American Academy of Political and Social Science, 415(1), 176-186. Glenn, N. D. (1980). Values, attitudes, and beliefs. In O. G. Brim & J. Kagan (Eds.), Constancy and change in human development (pp.596-640). Cambridge, MA: Harvard University Press. Goren, P. (2005).”Party identification and core political values”. American Journal of Political Science, 49(4), 881-896. Greenberg, E. S., Grunberg, L., & Daniel, K. (1996). Industrial Work and Political Participation: Beyond" Simple Spillover". Political research quarterly, 49(2), 305-330. Greenberg, Edward S. (1986). Workplace Democracy: The Political Effects of Participation. Ithaca,NY: Cornell University Press. Harold D. Clarke, Nitish Dutt and Allan Kornberg (1993). The Political Economy of Attitudes toward Polity and Society in Western European Democracies. The Journal of Politics, 55, pp 998- 1021. doi:10.2307/2131945. Heath, A., Evans, G., & Martin, J. (1994). The measurement of core beliefs and values: The development of balanced socialist/laissez faire and libertarian/authoritarian scales. British Journal of Political Science, 24(1), 115-132. 15 Jennings, M. K., Stoker, L., & Bowers, J. (2009). “Politics across generations: Family transmission re-examined”. Journal of Politics, 71(3), 782-799. Kjeldsen, A. M., & Jacobsen, C. B. (2012). Public Service Motivation and Employment Sector: Attraction or Socialization?. Journal of Public Administration Research and Theory. Klapper, J.T (1960). The Effects of Mass Communication. New York: The Free Press. Krosnick, J. A., & Alwin, D. F. (1989). Aging and susceptibility to attitude change. Journal of personality and social psychology, 57(3), 416-425. Krosnick, Jon A.; Alwin, Duane F. “Aging and susceptibility to attitude change” Journal of Personality and Social Psychology. Vol 57(3), Sep 1989, 416-425. Lewis-Beck, M. S., & Stegmaier, M. (2000).” Economic determinants of electoral outcomes.” Political Science, 3(1), 183. Lipset, Seymour Martin. 1959. "Some Social Requisites of Democracy: Economic Development and Political Legitima-cy." American Political Science Review 53:69-105. Mannheim, K. (1952). “The problem of generations”. In P. Kecskemeti (Ed.), Essays on the sociology of knowledge (pp. 276-322). London: Routiedge & Kegan Paul. March, James G., and Johan P. Olsen. 1995. Democratic governance. New York, NY: Free Press. Marini, M. M. (1987). “Measuring the process of role change during the transition to adulthood.” Social Science Research, 16(1), 1-38. Markus, G. B. (1979). “The political environment and the dynamics of public attitudes: A panel study”. American Journal of Political Science,23, 338-359. Michael Bratton and Robert Mattes (2001). Support for Democracy in Africa: Intrinsic or Instrumental?. British Journal of Political Science, 31, pp 447-474. doi:10.1017/S0007123401000175. Newcomb, T. M., Koenig, K. E., Hacks, R., & Warwick, D. P. (1967). Persistence and change: Bennington College and its students after 25 years. New York: Wiley. Newton, K., & Norris, P. (2000). Confidence in public institutions. Disaffected democracies. What's troubling the trilateral countries. Newton, K., & Norris, P. (2000). Confidence in public institutions. Disaffected democracies. What's troubling the trilateral countries. O'Donnell, Guillermo A. 1973. Modernization and Bureaucratic- Authoritarianism: Studies in South American Politics. Berkeley Pateman, Carol. (1970). Participation and Democratic Theory. Cambridge: Cambridge University Press. 16 Perry, James L., (2000). Bringing society in: Toward a theory of public-service motivation. Journal of PublicAdministration Research and Theory 10:471–88. Perry, James L., and Lois R. Wise. 1990. The motivational bases of public service. Public Administration Review 50:367–73. Robinson, Michael J. (1976). "Public Affairs Television and the Growth of Political Malaise: The Case of ’The Selling of the Pentagon."’ American Political Science Review 70(2):409-32. Schwartz Shalom H. (1994) Are there Universal Aspects in the Content and Structure of Values? Journal of Social Issues;50:19-45. Schwartz Shalom H. (2005) « Basic human values : their content and structure across countries » in A. Tamayo, J. B. Porto (eds.), Valores e comportamento nas organizações [Values and behavior in organizations], Petrópolis (Brazil), Vozes, pp. 21-55. Schwarz, N. (2007). “Attitude construction: Evaluation in context.” Social Cognition, 25(5), 638- 656. Scott, J. (1995). “Using household panels to study micro-social change. Innovation”: The European Journal of Social Science Research, 8(1), 61-73. Sears, D. O. (1975). “Political socialization”. In E I. Greenstein & N. W. Polsby (Ede.), Handbook of political science (pp. 96-136). Reading, MA: Addison-Wesley. Sears, D. O., & and Funk, C. L. (1999). “Evidence of the long-term persistence of adults' political predispositions”. Journal of Politics, 61, 1-28. Sears. D.O. (1975) “Political Socialization”. In F.I. Greenstein and N.W. Polsby, eds., Handbook of Political Science: Vol.2. Micropolitical Theory. Reading, Mass.: Addison-WesleyPp. 96-136 Shanahan, M. J. (2000). “Pathways to adulthood in changing societies: Variability and mechanisms in life course perspective”. Annual review of sociology, 667-692. Shaw, D. R. (1999). “The effect of TV ads and candidate appearances on statewide presidential votes, 1988-96”. American Political Science Review, 345-361. Sobol, Marion Gross. 1959. “Panel Mortality and Panel Bias,” in Journal of the American Statistical Association ,” 54:285 (Mar.) 52-68 Stevenson, B., & Wolfers, J. (2011). Trust in public institutions over the business cycle (No. w16891). National Bureau of Economic Research. Štulhofer, A., & Rimac, I. (2002). Opportunism, institutions and moral costs: the socio-cultural dimension of the underground economy in Croatia 1995-1999 (No. 14). Institute of Public Finance. Vaiman, V., Davidsson, P., & Sigurjonsson, T. (2010). Revising a Concept of Corruption as a Result of the Global Economic Crisis–The Case of Iceland. Organizational immunity to corruption: building theoretical and research foundations, Polish Academy of Sciences, 363-372. 17 van Deth, J. W., Abendschön, S., & Vollmar, M. (2011). Children and politics: An empirical reassessment of early political socialization. Political Psychology, 32(1), 147-174. Weatherford, M. S. (1992). “Measuring political legitimacy”. The American Political Science Review, 149-166. Wlezien, C., & Erikson, R. S. (2002). “The timeline of presidential election campaigns”. The Journal of Politics, 64(04), 969-993. Zaller, J. (1992). The nature and origins of mass opinion. Cambridge university press. 18 Figure 1: Aggregate Evolution of the duty to vote. Means by wave and age group. a) 16-30 years old b) 31-45 years old 19 Table 1: transition probabilities across categories and over time of duty to vote (xttrans). 16 to 30 years old Duty None Little Some A lot Total Duty None 74 4 16 7 100 Little 38 16 41 5 100 Some 24 6 41 29 100 A lot 9 1 21 69 100 Total 45 4 24 27 100 Table 2: transition probabilities across categories and over time of duty to vote (xttrans). 31 to 45 years old Duty None Little Some A lot Total Duty None 77 3 15 5 100 Little 35 13 41 11 100 Some 24 3 46 27 100 A lot 8 1 20 71 100 Total 45 3 24 28 100 20 Table 3: Between and within frequencies of duty to vote (xttab). 16 to 30 years overall Between within Freq, Percent Freq, Percent Percent Duty None 1423 46 650 62 75 Little 130 4 113 11 41 Some 738 24 179 46 52 A lot 783 26 389 37 66 Total 3074 100 1631 156 64 Table 4: Between and within frequencies of duty to vote (xttab). 31 to 45 years overall Between within Freq, Percent Freq, Percent Percent Duty None 2351 46 1064 60 78 Little 109 2 97 6 41 Some 1278 25 792 45 57 A lot 1358 27 668 38 68 Total 2096 100 2621 148 68 21 Table 5: Fixed-effect estimation of the evolution of duty to vote (Nov.2010-May 2012) 16-30 years old 31-45 years old Model 1 Model 2 Model 3 Model 4 Coef. Std. E. Coef. Std. E. Coef. Std. E. Coef. Std. E. Economic situation, working status Δ Income .35*** .16 .36*** .13 -.09 .14 -.08 .14 Δ Unemployed -.24 .16 -.22 .16 -.02 .1 -.02 .1 Δ Worker, permanent -.01 .08 -.00 .08 .06 .06 .06 .06 Economic perceptions Δ Eval.of the eco.sit (general) -.00 .13 .05 .1 Δ Eval.of the eco.sit.(retrospective) .09 .08 .06 .06 Δ Eval.of the eco.sit (prospective) -.05 .09 -.1 .06 Δ Eval.of the eco.sit (personal) -.13 .08 .01 .06 Δ Evaluation of the capacity of the PP -.03 .09 .06 .06 Δ Evaluation of the capacity of the PSOE .19 .12 .09 .1 Wave2 .09*** .04 .09*** .05 .06*** .03 .05*** .03 Wave3 .25*** .05 .27*** .05 .13*** .03 .14*** .04 Wave 4 .27*** .05 .29*** .08 .14*** .04 .14*** .04 constant 1.2*** .05 1.12*** .05 1.2*** .07 1.2*** .08 sigma_u 1.12 1.12 1.16 1.15 sigma_e .82 .81 .76 .76 rho .65 .66 .7 .7 F test that all u_i=0: F(2711) = 5.4 Prob. > F = .00 F(2711) = 5.3 Prob. > F = .00 F(2703) = 6.44 Prob. > F =.00 F(2703) = 6.27 Prob > F =.00 R2: within= .025 between= .0002 overall= .005 within= .028 between= .002 overall= .009 within= .01 between= .0003 overall= .0003 within= .01 between= .002 overall= .004 Corr(u_i, Xb) -.04 -.02 -.04 -.0003 Number of observations 3074 3074 5096 5096 Number of individuals 1043 1043 1773 1173 sigma_u: standard deviation of residuals within groups (individuals) ui sigma_e: standard deviation of residuals (overall error term) ei. Rho: % of rhe variance due to differences across panels, in this case, individuals ***= p<0.001, **= p<0.05, *= p<0.1 22 Annex 1: Summary of the descriptive statistics for the variables involved in the analyses. 14 The value zero gathers temporary workers, entrepreneurs, self-employed, member of a cooperative and “other”, including non-employed and non-active. Variable Obs. Mean Std. Dev. Min. Max. Duty to vote 8170 1.31 1.29 0 3 Income 8170 .38 .18 0 1 Work status: Unemployed 8170 .05 .23 0 1 Worker, permanent14 8170 .4 .49 0 1 Evaluation of PSOE to deal with the crisis 8170 .25 .25 0 1 Evaluation of PP to deal with the crisis 8170 .37 .25 0 1 Evaluation of the economic situation (gen.) 8170 .22 .2 0 1 Evaluation of the economic situation (retros.) 8170 .18 .28 0 1 Evaluation of the economic situation (prosp.) 8170 .49 .29 0 1 Evaluation of the economic situation (pers.l) 8170 .34 .33 0 1 Wave1 8170 .25 .44 0 1 Wave2 8170 .29 .46 0 1 Wave3 8170 .24 .43 0 1 Wave4 8170 .21 .41 0 1