Introduction: Socialization and Turnout

Abstention has been a major concern for electoral and political behaviour studies for half a century now. A large body of literature suggests that whether an individual is more or less prone to vote is, mostly, a matter of socialization (Hess and Torney, Reference Hess and Torney-Purta1967; Kiousis and McDevitt, Reference Kiousis and McDevitt2008; Linimon and Joslyn, Reference Linimon and Joslyn2002; Plutzer, Reference Plutzer2002; Verba et al., Reference Verba, Schlozman and Brady1995). Turnout is, however, conditioned by many other factors prior to the decision to vote, such as costs (for example, becoming informed about the choice), benefits and the probability that one's decision will decide the race (Mueller, Reference Mueller, Rowley and Schneider2003; Owen and Grofman, Reference Owen and Grofman1984). Whether and how individuals’ socialization will affect these elements of the vote utility function is not straightforward. However, most scholars include among these parameters a normative element, the “D term” or duty to vote (Blais, Reference Blais2000; Dowding, Reference Dowding2005; Knack and Kropf, Reference Knack and Kropf1998; Mueller, Reference Mueller, Rowley and Schneider2003; Riker and Ordeshook, Reference Riker and Ordeshook1968).

The belief that voting constitutes a duty is a moral attitude; it makes sense, then, to assume that it relies on individual's socialization—unlike the other determinants of turnout—yet very little research has been done on the causes and bases of the duty to vote (Blais and Galais, Reference Blais and Galais2016). For instance, Bowler and Donovan (Reference Bowler and Donovan2013) assumed that the differences they observed in the propensity to consider voting a duty stemmed from different socialization processes, as is the case for party identification or political interest. Nevertheless, the lack of appropriate data impeded them from putting their finger on the details of these specific “dutiful” socialization processes. As they concluded in their article, “for a fuller understanding of a sense of duty we may therefore need to develop theories of duty grounded in psychology or socialization in order to develop a baseline level of civic duty for a given individual” (Bowler and Donovan, Reference Bowler and Donovan2013: 272).

From a traditional perspective, the belief that voting is something good while abstaining is bad reflects a moral conviction that in turn stems from a social norm about what a good citizen should do (Blais, Reference Blais2000; Dalton, Reference Dalton2008; Knack and Kropf, Reference Knack and Kropf1998; Mueller, Reference Mueller, Rowley and Schneider2003; Uhlaner, Reference Uhlaner1986; Zuckerman and Kotler-Berkowitz, Reference Zuckerman, Kotler-Berkowitz and Swaine1998). Hence, we should explore the role of the socialization agencies that play a role in the process of internalization of such social norms, namely, the school and the family. Nevertheless, the literature on political socialization suffers from a series of limitations, including a scarcity of longitudinal data and a tendency to deal with only one factor or agency at a time, thus hampering the correct attribution of explanatory power to each phenomenon. This research strand has also disregarded specific dynamics within the family and the school and the theoretical contributions of other disciplines, such as developmental psychology, that could shed light on the causal mechanisms of some well-known relationships. This article seeks to fill these gaps in the literature regarding the socialization processes and electoral participation by focusing on the early dynamics and agencies that result in dutiful adult citizens.

The article is structured as follows. First, it explores the literature on the civic duty to vote and its link with early socialization processes. Second, I explain the research design and the dataset used to explore the socialization pathways to duty, the National Longitudinal Survey of Children and Youth (NLSCY), a Canadian panel covering the 14 years between 1994 and 2008. This dataset allows for the testing of the relationship between childhood features and dynamics measured when the children were between five and 12 years old and the duty to vote, which was measured when they were between 19 and 26 years old. Third, the article presents the results of testing the role of three familial factors (socioeconomic status (SES), parenting styles and parental engagement with children's education) and two school features (a school's democratic governance and faith-based schools) in relation to duty. The article concludes with a discussion of this research's main findings.

Why Study Socialization and Duty?

The belief that voting is a citizen's duty has been a major predictor of turnout since Campbell and colleagues noted in Reference Campbell, Converse, Miller and Donald1960 that turnout was 70 percentage points higher among those with a strong sense of duty than among those with none (105-06). Other scholars have agreed that this belief is crucial for most citizens when it comes to deciding whether or not to attend the polls (Blais, Reference Blais2000; Riker and Ordershook, Reference Riker and Ordeshook1968; Verba et al., Reference Verba, Schlozman and Brady1995). Among the few authors concerned with the origin and bases of this decisive attitude, Klemmensen and colleagues stated that “socialization, acculturation and other environmental factors are still the best candidates when explaining variations in civic duty” (Reference Klemmensen, Hatemi, Hobolt, Petersen, Skytthe and Nørgaard2012: 214). Yet, to date, there has been no systematic research on how the sense of civic duty is instilled or enforced through the socialization process; there is a gap in our understanding of where a given individual gets his sense of civic duty (Bowler and Donovan, Reference Bowler and Donovan2013).

Political socialization can be defined as the set of processes by which society inculcates citizenship norms, transmitting political culture from one generation to another (Almond, Reference Almond1960; Almond and Verba, Reference Almond and Verba1963; Hyman, Reference Hyman1959). Since its first steps, this literature has switched its attention from the initial stages of the socialization process (Greestein, Reference Greenstein1965; Hess and Torney, Reference Hess and Torney-Purta1967) to later phases of the life cycle (namely adolescence, young adulthood and even later), driven by the idea that politics was a subject of adult life and that the complexity of political attitudes makes such attitudes tricky to measure in individuals under the age of fourteen (Niemi and Hepburn, Reference Niemi and Hepburn1995). This switch in focus decreased attention to the role of parents (Achen, Reference Achen2002; Amna et al., Reference Amnå, Ekström, Kerr and Stattin2009), and has abandoned the idea that some values and attitudes were learned early and remained stable through the life cycle (Searing et al., Reference Searing, Schwartz and Lind1973).

Yet there are reasons to expect that the bases of the duty to vote form early in life. Studies about general, symbolic orientations, particularly those related to patriotism and views on citizenship, proved that these are not only formed early (Sears et al., Reference Sears, Hensler and Speer1979), but also that they correlate with a number of policy and party preferences expressed later in life (Sullivan et al., Reference Sullivan, Fried and Dietz1992; Theiss-Morse, Reference Theiss-Morse1993). Recent research has found that children as young as five-and-a-half years old already have an approximate idea about the norms of citizenship (Van Deth et al., Reference Van Deth, Abendschön and Vollmar2011).Footnote ¹ Similarly, dispositions toward electoral activity have been found to be formed even earlier (Keating and Janmaat, Reference Keating and Janmaat2016). It makes sense, then, to turn our attention to the early socialization dynamics that pave the way to dutiful citizenship.

Aside from a lack of interest in the sense of duty to vote in general, and a more specific inattention towards the infancy processes that make dutiful adults, the political socialization literature still faces some challenges that the present research aims to overcome. First, there is only limited evidence based on a longitudinal perspective that helps us understanding how and when citizenship norms are formed (Amna et al., Reference Amnå, Ekström, Kerr and Stattin2009; Niemi and Hepburn, Reference Niemi and Hepburn1995; Owen, Reference Owen2008; Sapiro, Reference Sapiro2004). Second, most studies consider only one agency or context at a time instead of examining several socialization factors and their interplay, thus not allowing us to effectively isolate and attribute a particular effect to each factor (Amna et al., Reference Amnå, Ekström, Kerr and Stattin2009). In relation to this shortcoming, many studies limit themselves to showing differences between types of families or schools, but “little is known about the dynamics underpinning the political socialization process” (Owen, Reference Owen2008: 16) which highlights the need to unravel the specific courses of action that convey the sense of civic duty. Hence, a clearer focus on the dynamics of socialization is needed. Last but not least, there is a deficit of interdisciplinary dialogue with other research fields that limits the explanatory power of political socialization studies. More specifically, several scholars strongly suggest that developmental psychology can provide a better understanding of the childhood factors behind general political attitudes (Amna et al., Reference Amnå, Ekström, Kerr and Stattin2009, Bowler and Donovan, Reference Bowler and Donovan2013). In the words of Sapiro, “there is room for reinvigorating the connections between developmental psychology and political socialization” (Reference Sapiro2004:14). Borrowing from developmental psychology the concepts and dynamics that would result in dutiful grown-ups, this article contributes to political socialization studies not only by paying attention to the sense of duty and its early roots, but also by adopting a longitudinal perspective, and systematically analyzing the effects of the primary agencies of socialization operating during infancy.

Family, School and Duty

Family is the first—and for some, foremost—socialization agency in life. Family's influence on voting propensity and other forms of participation has been confirmed several times (Hess and Torney, Reference Hess and Torney-Purta1967; Verba et al., Reference Verba, Schlozman, Burns and Zuckerman2005). Parents can directly affect the future behaviour of their children by providing an example of engagement and, for this study in particular, the act of voting (Plutzer, Reference Plutzer2002; Tedin, Reference Tedin1974). Nevertheless, the relationship between parents’ and children's political involvement has been found to be entirely mediated by political attitudes such as interest in politics (Quintelier, Reference Quintelier2014). Hence, it is safe to assume that any family effect on turnout will be mediated by the most important subjective factor preceding turnout: duty.

Although there might be some specific rules about the duty to vote within the household, it makes more sense to expect this symbolic attitude to emerge as a result of adults emphasizing attachment and obedience in their earliest interactions with children (Sapiro, Reference Sapiro2004). This indirect path would be more closely related to family dynamics than to specific messages, in the same way that discussing politics is a well-known predictor of critical and politicized attitudes (Gimpel et al., Reference Gimpel, Celeste Lay and Schuknecht2003; McIntosh et al., Reference McIntosh, Hart and Youniss2007; Verba et al., Reference Verba, Schlozman and Brady1995). In this respect, developmental psychology makes a valuable contribution by describing how parents deliver discipline to children (Baumrind, Reference Baumrind1991).

The theory employed in this article organizes parenting styles into two dimensions: control and warmth. This results in four styles: authoritative (high in both dimensions), authoritarian (high in control, low in warmth), permissive (low in control, high in warmth), and neglecting (low in both dimensions). The style most closely associated with political attitudes related to participation seems to be the authoritarian parenting style. It is characterized by parents being high in control and low in warmth. As this authoritarian style is more frequent in politically alienated parents and results more frequently in less-open-to-experiences offspring (Peterson et al., Reference Peterson, Smirles and Wentworth1997), it should be negatively correlated with the duty to vote. Another notion used in parenting studies theory is consistency, parents administering the same cues and discipline through the whole of the children's socialization. Consistency is believed to sustain children's cognitive skills (Landry et al., Reference Landry, Smith, Swank, Assel and Vellet2001) and ease children's assimilation of norms; thus, it might also help in the development of the sense of the duty to vote.

Another salient factor in the developmental psychology literature is parental engagement with children's education. This phenomenon is multidimensional, and encompasses engagement, responsibility and accessibility. While the first subdimension is defined as “time spent in actual one-on-one interaction with the child” (Lamb, Reference Lamb1986: 8), accessibility entails availability to the child on a daily basis (without the need of direct involvement). Finally, responsibility means being accountable for the child's welfare (such as arranging and attending appointments, dressing and feeding, making childcare arrangements, buying the child's necessities for his education). Parental engagement with a child's education has been related to a more internal locus of control of the child (which is very close conceptually to sense of duty), and a better social-relational functioning later in life, aside from children's academic achievement and other positive cognitive and psychological effects (Sarkadi et al., Reference Sarkadi, Kristiansson, Oberklaid and Bremberg2008). Although this literature does not connect parental engagement and the child's political participation or political attitudes, the link with civic duty, particularly in regard to the “responsibility” sub-dimension, seems pretty straightforward. By fulfilling their obligations with the child's schooling, parents are “preaching by practice.” Showing their involvement with the school community, the child might learn to do so in the future with a broader community. Doing what is right and expected from parents teaches the child to do the same in due time. Therefore, the children whose family is involved in their education should be more dutiful in the future than those raised by more disengaged families.

Because “socio-economic variables often confound both parenting behaviours and child outcomes” (Sarkadi et al., Reference Sarkadi, Kristiansson, Oberklaid and Bremberg2008: 154), work on the causes and effects of parental involvement in children education almost always includes socioeconomic status (SES) controls. In turn, high levels of education or income are related to high likelihood of voting (Jennings et al., Reference Jennings, Stoker and Bower2009). Wealthier families provide more and better resources necessary to develop civic skills, such as the habit of reading newspapers (Verba et al., Reference Verba, Schlozman and Brady1995). Conversely, economic constraints may divert parents’ interests from politics and reduce the quality and quantity of parent-children interaction, thus hampering the transmission of attitudes and skills that may spur children's future sense of duty (Pacheco and Plutzer, Reference Pacheco and Plutzer2008). The effect of family SES is expected to diminish through the life cycle as experience compensates for the lack of other resources early in life (Wolfinger and Rosenstone, Reference Wolfinger and Rosenstone1980), and also by the effect of other levelling institutions that mitigate the effect of family SES (Pacheco and Plutzer, Reference Pacheco and Plutzer2008), such as school.

The most obvious causal mechanism linking school and civic duty is the acquisition through schooling of civic skills and attitudes known for boosting the likelihood to vote, such as political knowledge (see Delli et al., Reference Delli Carpini and Keeter1996, Gimpel et al., Reference Gimpel, Celeste Lay and Schuknecht2003; Verba et al., Reference Verba, Schlozman and Brady1995). From the perspective of social reproduction theory, the primary function of schools is to reproduce the social structure and organization of society (Bourdieu, Reference Bourdieu, Karabel and Halsey1977), maintaining the social order and transmitting the dominant social norms. Following this train of thought, school socialization perpetuates the norms consistent with the governance of the nation (Dennis, Reference Dennis1968).

Newer perspectives, however, focus on school dynamics as an indirect source of civic attitudes and behaviours. Some scholars have proved the effects of open classroom climates—that is, the settings that offer opportunities to students to form and express an independent opinion—on free thinking, political discussion, civic attitudes and political engagement (Flanaganet al., Reference Flanagan, Bowes, Jonsson, Csapo and Sheblanova1998; Torney-Purta et al., Reference Torney-Purta, Lehmann, Oswald and Schulz2001). A less explored facet of school dynamics likely affecting civic duty is school governance. In practice, schools that allow students to determine relevant aspects of teaching, resource allocation or class dynamics are considered to be more “democratic,” but so are those offering this opportunity to teachers. The more democratic the school governance is, the higher the levels of political engagement among its students (Mosher, Reference Mosher1994; Pasek et al., Reference Pasek, Feldman, Romer and Jamieson2008). Intuitively, we can expect that a school whose staff is allowed and encouraged to take part in the decision-making process conveys the value that an active role in such processes is important and that everyone in a given community must fulfill their obligations for the sake of a greater good.

An attempt to integrate the three aspects of horizontal school dynamics is offered by Finn (Reference Finn1989). According to Finn's participation-identification model, students’ active role in class precedes their participation in school activities outside of the formal curriculum, which in turn is a stage previous to students’ participation in school governance. Each of these “horizontal” traits of school dynamics will have a positive effect on students’ performance, sense of self-efficacy, and identification with the school community. We can derive two conclusions from Finn's model: first, a school's democratic governance may indicate that all other traits of horizontal dynamics are present, too; second, these horizontal dynamics not only engage children in political participation, but also generate a sense of belonging to a community that paves the way for the assimilation of social norms such as the civic duty to vote.

Indeed, schools are the first environment to enable agency after a student's family, and they satisfy the psychological need of belonging to a community (Battistich et al., Reference Battistich, Solomon, Kim, Watson and Schaps1995; Osterman, Reference Osterman2000), providing a primary sense of belonging. Not only “horizontal” schools do better in building this parameter, religious schools seem to have an advantage in this respect as well (Burtonwood, Reference Burtonwood2003; MacMullen, Reference MacMullen2004). Although religious schools may not convey autonomy or social cohesion as much as a sense of belonging to a social community, the latter is considered a necessary condition for acquiring and internalizing social norms. Moreover, the language of duties is very germane to religious thinking (Macaluso and Wanat, Reference Macaluso and Wanat1979); and both religiousness and duty belong to the moral domain and the sub-domain of authority (Graham et al., Reference Graham, Nosek, Haidt, Iyer, Koleva and Ditto2011). It makes sense, therefore, that religious schools would nurture the idea that children have duties to perform, among them voting.Footnote ² Hence, religious schools might do better than secular schools in instilling a sense of duty in children.

This article addresses the socialization processes that lead to dutiful adult citizens or, more specifically, the aspects and dynamics of the family or the school that contribute to the sense of duty during early childhood. The literature reviewed above overcomes some of the characteristic deficits of the political socialization literature. The general expectations of this research are the following: a) familial influence is expected to be strong and conveyed through three mechanisms: parenting styles, parents’ engagement with children's education and socioeconomic status; b) school is expected to positively affect duty—and to lessen the effects of family SES—through two mechanisms: democratic governance and religious (and therefore, duty-based) values.

Research Design and Variables

In order to test the expectations of this research, a survey tapping socialization processes was needed. The NLSCY is an eight-cycle panel study, which follows a cohort of Canadian children from birth to early adulthood, between cycle 1 (1994-1995) and cycle 8 (2008-2009).Footnote ³

Canada makes an interesting case study because it is one of the most dutiful countries in the world. The Making Electoral Democracy Work surveys (2011-2015) show that Canada has the greatest share of dutiful citizens among the five participant countries: 70 per cent of Canadians claim to have a duty to vote, while only 32 per cent of Swiss or 27 per cent of German citizens believe that they have a duty to vote in elections. Nevertheless, Canada has exhibited a downward pattern in its turnout rates in recent years that has been blamed on the fact that younger people are less likely than the older to construe voting in terms of a moral duty (Blais and Rubenson, Reference Blais and Rubenson2013; Blais et al., Reference Blais, Gidengil and Nevitte2004). Probably out of this concern, the very last wave of the NLSCY survey, released in 2011, included questions about feelings towards voting for those between 19-20 and 25-26 years old.Footnote ⁴ Among these questions, the ones for which we have also measures of their socialization processes at cycle 1 (5-6 to 11-12 years old at that time) constitute our sample.

Hence, the following analyses are restricted to the individuals that answered a question about their duty to vote in the last wave of the NLSCY study and who were in the study from the very beginning. This makes a total of 4,299 individuals.Footnote ⁵ Of note, the study tracks a child's cohort during the totality of their “impressionable years.” Indeed, the convention is to consider that the impressionable years—that is, the age period at which an individual is open to the influence of socialization agencies and context—begin in adolescence and end in the early years of adulthood. Depending on the author, the process can start as early as 11 years old and end at age 25 as in the latest (Krosnick and Alwin, Reference Krosnick and Alwin1989).

The last wave of the NLSCY survey included two versions of the Blais/Achen question (see Achen and Blais, Reference Blais and Achen2010) on the duty to vote. These questions are: “Generally speaking, do you believe that you have a duty to vote in every federal election?” and “Do you believe that you have a duty to vote in every provincial election?” The available answers were yes (1) and no (0). The questions were recoded into a new dichotomous variable, which uses the value 1 if the individual answered “yes” in both cases and 0 if otherwise.Footnote ⁶

As for the measure of the independent variables, we must first introduce the structure of the survey. In the first wave of the study, children between 5 and 6 and 11 and 12 years old only answered vocabulary and mathematical aptitude questionnaires, which are of no use in this research. A parents’ questionnaire was available for all children and a teacher's questionnaire was available for some. The parents’ questionnaire was filled out by the person most knowledgeable about the children in the household, the mother most of the time. This person provided information about herself, her partner and the children who were tracked by the survey. If the most knowledgeable person gave parental consent for collecting school information, questionnaires were mailed to the schools and filled in by teachers. This design complicates the database structure in the sense that we have measures regarding school variables (that is, the type of school) that are provided by parents, and measures regarding the family (that is, whether or not children's parents engage in school activities) that are provided by a teacher, and only for some children.Footnote ⁷ We must therefore be cautious when interpreting the effect of a variable that might reflect the subjective perception of a teacher on familial influence. Moreover, the inclusion of variables measured using the teachers’ questionnaire further restricted the sample to 1,695 individuals. Nevertheless, this structure provides a double protection against endogeneity. First, the independent variables were measured long before the dependent variable; second, they do not reflect children's subjectivity but parents’ and teachers’ perceptions.

The parents’ questionnaire included indicators of family SES, parenting style and a question on the kind of school their children attended. The family SES indicator gathers information from five questions on total years of schooling for both parents, occupational status of both parents and the household income in thousands of dollars. The variable has been rescaled so that 0 indicates the minimum possible social status in the dataset, and 1 indicates the highest possible social status. Parenting styles were measured using two scales based on Dr. M. Boyle's measures, in turn adapted from Strayhorn and Weidman's Parent Practices Scale. The authoritarian factor is originally named “punitive or aversive factor” and built using four questions on how often children were punished when they broke rules (Cronbach's alpha = .6). A consistent parenting style is gauged by means of a scale using five questions on the frequency with which parents fulfilled their threats of punishment in case of misbehaviour (Cronbach's alpha = .7). Finally, a question answered by the parents identified whether the school the child was attending was a public school, a publicly funded Catholic school, a private school or other. Since private schools were a minority (with only 351 children from the pool of participants), this variable was collapsed, with the value 1 identifying Catholic schools, and 0 others.

The teachers’ questionnaire included information for each child about their parents’ engagement with the school. This factor variable was built using five questions that asked how prepared the students were for school with regards to school materials and clothing, and whether the child managed to complete school assignments. Higher values indicate higher parental engagement. Teachers also provided two aggregate measures: the percentage of families of low income in the school and a democratic governance index. While the first measure taps the school's SES, the second is a factor built using seven questions about the “social climate of the school” which were agree/disagree statements about the degree of horizontality in decision making. Higher values indicate more involvement of the staff and teachers in school policies and in how the resources are allocated.

Finally, a series of temperament indicators were included as controls. Recent works have highlighted the role of personality on the development of sense of duty (Blais and Labbé St-Vincent, Reference Blais and St-Vincent2011; Weinschenk, Reference Weinschenk2014). An important dimension of personality is temperament, that is, the individual's propensity to react and self-regulate in a particular way (Rothbart and Derryberry, Reference Rothbart, Derryberry, Lamb and Brown1981). These indicators were included also because a recent research strand suggests that duty has a genetic origin (Klemmensen et al., Reference Klemmensen, Hatemi, Hobolt, Petersen, Skytthe and Nørgaard2012; Loewen and Dawes, Reference Loewen and Dawes2012), such as other pro-social attitudes related to electoral participation (Fowler et al., Reference Fowler and Dawes2008). Given that temperament is a precursor of personality that arises from our genetic endowment (Rothbart et al., Reference Rothbart, Ahadi and Evans2000), the three temperament aspects included in the questionnaire for five- to-12-year-olds were considered in the models: aggression, hyperactivity and pro-social disposition.Footnote ⁸ An additional pro-social behaviour indicator was provided by the teacher in their questionnaire, when asked about the children's social skills at school. The scale gathers information from six questions on the degree of co-operativeness, respect and self-control of each child.Footnote ⁹

Figure 1 displays an overview of the causal model proposed in this research. The figure clarifies the main factors whose influences were empirically tested, the indicators used for that purpose, as well as the controls taken into account. It also offers insight into the time in which they were measured, in the sense that the independent variables measured in T1 (cycle 1 of the study) were expected to have an effect on duty measured in T2 (cycle 8 of the study), 14 years later. Note that age is controlled by keeping the cohort constant, and that children's age (which barely varies) is not included in the estimations.Footnote ¹⁰

Figure 1 Causal schema, main variables and timing of the research design.

The measure of the dependent variable determines the estimation method chosen. Hence, whether or not individuals feel a sense of civic duty (age 19-26) will be estimated by means of logistic regressions. All variables have been recoded in a range between 0 and 1; hence the coefficients may be interpreted as the effect of a variable varying from its minimum to its maximum value.