Introduction

Democratic societies require that every citizen gets an equal opportunity to influence political decisions (Dahl, 2006). However, it is a robust finding that some groups participate structurally more than others and that socioeconomic background predicts political engagement well (Schlozman et al., 2012). Bovens and Wille (2017) argue that we live in a diploma democracy where those with high educational attainment are more likely to take an active and effective political role. After all, they can rely on resources such as knowledge, skills, time, money, and connections necessary for political participation (Brady et al., 1995). Paralleling their older counterparts, young people with low educational attainment engage less with politics than highly educated youngsters (Schlozman et al., 2010). Next to the general concern about low levels of youngsters’ political engagement compared to older cohorts (Sloam, 2013), scholars call for more attention to these intra-generational inequalities in addition to the inter-generational differences in political engagement (Grasso & Giugni, 2022).

In response to this lacuna in the literature, we study socioeconomic inequalities in internal political efficacy (IPE). More specifically, we conceptualize and operationalize the concept of a political home environment that we embed within Bourdieu and Passeron’s (1990) general reproduction theory. We empirically test its effect on feelings of IPE, an important predictor of future political participation (Levy & Akiva, 2019). Since we study the political behavior of senior high school students who do not yet have voting rights, studying inequality in their IPE provides insights into their future political behavior before they enter the political arena.

We test whether citizenship education affects the inequality caused by youngsters’ political home environment in feelings of IPE. Since not every child grows up in a stimulating political home environment, many hope that citizenship education can compensate for the lack thereof. In fact, previous studies have shown that citizenship education can foster civic engagement and compensate for the lack of a stimulating political home environment (Campbell, 2019; Neundorf et al., 2016). However, these effects are not as strong for each type of citizenship education and each group of students (Hooghe, 2012). For instance, Hoskins and Janmaat (2019) show that citizenship education can have a differential effect on children from diverging social backgrounds. In other words, the effect can take different directions. Although a compensation effect is desired, inequalities can also be reproduced or accelerated due to citizenship education at school.

Moreover, we study whether citizenship education as a potential equalizer for those not growing up in a stimulating political home environment is affected by their educational track. After all, previous research has found that educational tracking, as a characteristic of the educational landscape, negatively affects feelings of stigmatization, educational achievements, and future political participation (Hoskins et al., 2016; Spruyt et al., 2015). Others have shown that vocational students receive fewer quality citizenship education opportunities (Hoskins & Janmaat, 2019; Nieuwelink et al., 2019). We build on these studies by testing how educational tracking influences students’ IPE. At the same time, we also go beyond the extant literature by studying how citizenship education as a potential equalizer for students’ political home environment differs across educational tracks. The rationale is that educational environments construct “class-based identities of active or passive citizens” (Willeck & Mendelberg, 2022, p. 94), which might influence students’ IPE levels. In such an educational context, students in the technical and vocational tracks are more likely to be socialized that politics is not for their kind of students after being exposed to citizenship education. After all, it is likely that citizenship education will at least implicitly echo the widely spread idea that politics is an activity primarily for and by the higher educated in our diploma democracies (Bovens & Wille, 2017).

In summary, this paper will investigate whether three citizenship education components at school compensate, reproduce, or accelerate inequalities in students’ IPE and how this relationship differs across educational tracks. We study the effect of students’ perceived civic learning experiences, open classroom climate for discussion, and active student participation at school. IPE is especially interesting considering the stigmatizing effects of educational tracks since it taps into self-confidence in a political setting. Based on cross-sectional data collected to test the attainment targets of citizenship education in the twelfth grade in Flanders, we conducted a correlational study with three-way interactions in multilevel models. The Flemish school system is an often-studied case of a school system that is highly differentiated through early tracking (Vandenbroeck et al., 2022). Findings from the Flemish case are relevant not only to similar European countries in which early tracking is implemented, like Germany, Austria, and the Netherlands, but also to the lively international debate about the merits and deficiencies of tracking (OECD, 2012).

Political Home Environment

One’s family is considered a main, if not the most important, socialization agent in transmitting political engagement (Jennings et al., 2009; Neundorf & Smets, 2017). Parents influence their children’s political development either directly or indirectly. The direct intergenerational transmission of political engagement refers to social learning theory, which states that children learn through observing and imitating (Bandura, 1977a). Parents function herein as role models. Politically active parents open the political world to their children, who observe and imitate their parents’ behavior. This political home environment stimulates these children and affects their adult life. Empirical data supports this claim by showing significant relationships between parents and their offspring regarding political participation and political attitudes, such as their levels of political trust, interest, and efficacy (Jennings, 2009; Quintelier, 2015).

Parents’ indirect way of influencing their children’s political socialization process refers to their family characteristics contributing to a fertile social milieu for political engagement. This social milieu pathway or status transmission model posits the inheritance of parents’ social status accompanied by its level of political engagement (Neundorf & Smets, 2017; Quintelier, 2015). The relationship between economic, social, and especially cultural capital and political engagement is robust (Brady et al., 2015; Schlozman et al., 2012). Hence, parents transmit the resources needed for political engagement to their children. Children from highly educated parents are more likely to become highly educated themselves, influencing their political engagement.

This direct and indirect intergenerational transmission of political engagement fits with Bourdieu and Passeron’s general reproduction theory, which explains the intergenerational transmission of cultural capital and its accompanying status (Bourdieu, 1989; Bourdieu & Passeron, 1990). In analogy with the distinctions made above, cultural capital consists of objectified, institutionalized, and embodied cultural capital (Kraaykamp & van Eijck, 2010). Correspondingly, this refers to cultural objects such as books, educational credentials such as a diploma, and long-lasting dispositions of the mind and body, such as how somebody acts. Especially this last form of cultural capital is closely intertwined with habitus, the system of dispositions mediating between the objective structure and practices. Bourdieu (1989, p.171) calls this “a structured and structuring structure” that consists of a system of schemes wherein thoughts, observations, evaluations, and intentions are embedded. This Structure-Disposition-Practice (SDP) scheme explains why socioeconomic differences at an individual level impact behavior (Bourdieu, 1977). These dispositions are first and foremost transferred by the family and secondly by the school, peers, and other socialization agents. The resulting internalized habitus generates meaningful social practices and perceptions fitting the original social context (Kraaykamp & van Eijck, 2010; Nash, 2003).

In Bourdieu’s terminology, this paper studies how parents’ capital in the political field nurtures a political habitus trickling down to their children. Regarding the transmission of political engagement, children growing up with parents who talk a lot about politics and show an interest in politics (embodied political capital), who have many cultural goods such as books (objectified cultural capital), and who have high educational credentials (institutionalized cultural capital) are expected to form a political habitus which combines the direct and indirect transmission of political engagement. Parents’ political habitus provides their children with resources and experiences, resulting in knowledge, skills, and a sense of political entitlement. This sense of entitlement gives them the feeling that they can express their opinions publicly, whereas a lack thereof results in self-exclusion (Bourdieu, 1989). This highlights the need to look beyond the cost and skill-based explanation for inequality in political participation (Laurison, 2015). Inequality in political participation is not only a matter of knowledge and skills; citizens must also believe they are legitimate political participants.

Closing the Inequality Gap with Citizenship Education

Scholars and policymakers believe schools can foster active citizenship and compensate for lacking politically stimulating environments. While Langton and Jennings’(1968) early study showed no influence of civic education on political attitudes, several studies conducted over the last decades have contested their findings, arguing that civic education can foster active citizenship (Campbell, 2019; Niemi & Junn, 1998). Positive effects of citizenship education on civic dispositions, such as IPE, are empirically confirmed (Campbell, 2019; Hoskins et al., 2017; Niemi & Junn, 1998; Pasek et al., 2008). Nevertheless, citizenship education does not affect every group equally (Neundorf et al., 2016). Some groups might gain more from citizenship education than others.

We can expect three potential moderation effects of citizenship education on social class inequalities in political engagement in line with previous scholarship (Hoskins et al., 2021; Neundorf et al., 2016). Citizenship education can compensate, reproduce, or accelerate inequalities in political engagement, including IPE. Compensation is the most favorable moderation effect for promoting equality in IPE. This indicates that less privileged children gain more from citizenship education than privileged ones. A compensation effect occurs when children with less political exposure at home benefit more from initial exposure to politics, leading to a catch-up effect. After all, when a student has a very low starting point, there is more room for growth. In contrast, students with a higher starting point reach a ceiling, and it becomes more difficult to grow. An impressive range of studies shows how well-designed citizenship education compensates for a disadvantaged background (Campbell, 2019; Deimel et al., 2020; Hoskins & Janmaat, 2019; Neundorf et al., 2016). These studies point to the potential of citizenship education to help students from disadvantaged backgrounds catch up and close the political inequalities gap.

However, inequalities in political engagement might also be reproduced or accelerated due to citizenship education. This happens when the gap between privileged and less privileged children persists after an intervention or when it enlarges (Neundorf et al., 2016). In the latter scenario, privileged children’s political engagement increases more than their less privileged peers. This might be because privileged students “have experience and skills on which to build” (Hoskins et al., 2021, p. 96). These children from families with high cultural capital might also feel more comfortable in a school environment that aligns with their upbringing and expectations, whereas those who are not may struggle to acquire new knowledge or skills, leading to reduced self-confidence. As a result, students may become more entrenched in their feelings of disentitlement (Bourdieu & Passeron, 1990). Some empirical studies point to a potential reproduction or even acceleration of inequalities. They highlight how different effects can emerge in different contexts. Mulder (2021), for instance, tempers the optimism about citizenship education as a panacea for inequalities as her results suggest that a Dutch on-site citizenship program did not reduce pre-existing inequalities (see also Persson, 2012). Hooghe and Dassonneville (2011) even find that the inequality gap became more pronounced and accelerated after civic learning.

Given these mixed results, it is unclear why citizenship education sometimes brings about compensatory effects on political engagement outcomes while others show reproduction or acceleration effects. One potential explanation is that the moderating effects of citizenship education vary depending on the learning strategy and associated learning goals. Children with differing levels of prior political exposure may respond differently to various learning strategies. Citizenship education can generally be divided into two types. The first type mainly aims to transfer political knowledge and establish a foundation in which the student is viewed primarily as a recipient of information. The second type employs active learning strategies, such as having political discussions or participating in student councils, in which the student actively participates (Hoskins & Janmaat, 2019, p. 16). While active learning strategies are shown to be the most effective in increasing overall political engagement (Willeck & Mendelberg, 2022), this might be different for alleviating inequalities in political engagement. Active learning strategies may be better suited to students who already possess political knowledge and skills, while those from less privileged backgrounds may struggle to catch up and experience a “left behind” process (Neundorf et al., 2016, p. 927). Conversely, laying a foundation through traditional civic learning opportunities may benefit students with less political exposure, as it can level the playing field.

Schools Structuring Inequality

Alongside the school’s function as a political socialization agent, schools have an allocation function by structuring children based on their achievements. In Durkheimian terms, “schools differentiate and assign children a place in the system of social stratification” (Kavadias et al., 2017, p. 31). The school system establishes this by evaluating students and differentiating them into tracks based on their academic ability. Supporters argue that tracking creates learning environments adapted to students’ talents, whereas critics show how it reproduces inequalities (Ballantine et al., 2017). Children with a low socioeconomic status have a higher chance of being allocated to vocational tracks, and children growing up with much cultural capital have an advantage (Bourdieu & Passeron, 1990). Moreover, tracking enlarges this advantage as research shows that children in higher tracks experience more and better learning opportunities, causing more learning gains (Ballantine et al., 2017). Hence, less privileged children are not only more likely to be allocated to vocational tracks, but tracking itself further reinforces these inequalities. We study whether this also holds for citizenship education and if citizenship education’s potential to be a democratic equalizer differs by educational track.

A burgeoning body of literature shows how educational tracking inhibits political learning and reproduces pre-existing inequalities in political dispositions (Hoskins et al., 2016; Janmaat et al., 2014; van de Werfhorst, 2009). The academic segregation where children with a low socioeconomic background end up disproportionally more in vocational tracks translates into a group of children from lower politically stimulating home environments (cf. supra). Alongside this disadvantaged starting position, research shows that students in vocational tracks get less and less effective citizenship education. While students in academic tracks are regularly encouraged to discuss and engage in political activities at schools, this is much less the case in vocational tracks (Hoskins & Janmaat, 2019; Nieuwelink et al., 2019).

Besides a direct effect of tracking on IPE, we might expect the moderating effect of citizenship education on the relationship between one’s political home environment and IPE to differ across tracks. Cultural theories of class-based inequalities in political participation argue that educational environments construct “class-based identities of active or passive citizens” (Willeck & Mendelberg, 2022, p. 94). Inspired by Bourdieu (1989), they contend that social arenas such as schools create class cultures that reproduce different tastes towards politics. The taste or intention to engage with politics is reproduced as an activity for middle- and upper-class citizens. Students are socialized to believe that those dominant groups in society belong in the political arena and others do not. Their relationship with politics differs based on their social background and educational track. Students know the prestige associated with a track, and an adolescent’s future social position highly depends upon the educational track they are sorted in (Batruch et al., 2019; Boone & Demanet, 2020; Spruyt et al., 2015).

Therefore, we can expect that when children are taught about politics, they are indirectly socialized about whether politics is a suitable interest and activity for students within their class. After all, in our contemporary diploma democracy, politics is constantly depicted as an activity in which highly educated citizens are legitimately overrepresented (Bovens & Wille, 2017). This might cause differential political learning experiences for students from different tracks. Whereas students from the academic track indirectly learn that politics is suited for them as future higher-educated citizens, vocational students might indirectly learn the opposite due to citizenship education. Students in vocational and technical tracks might not only receive less qualitative citizenship education (Nieuwelink et al., 2019), but citizenship education might also foster a class-based identity and corresponding taste that politics is not for their kind of student (Bourdieu, 1989). Meanwhile, students from the academic track with less politically stimulating home environments might start believing they can climb the political-social ladder and develop an upper-class-based identity since they are on their way to becoming higher-educated citizens fitting the diploma democracy.

I (Do Not) Feel Confident About Participating in Politics

We study citizenship education’s possible compensation, reproduction, and acceleration effects on IPE. This political attitude refers to the belief in one’s ability to influence political decisions and corresponds with Bourdieu’s concept of political competence (Bourdieu, 1989). To understand why people become politically engaged, it is not sufficient to consider someone’s competencies. We also need to consider a person’s perception of their competence. IPE conceptualizes someone’s feelings of ability in the political field, but this assessment is established in dialogue with one’s environment. Referring to the SDP scheme, we conceive IPE as a disposition shaped at home and potentially influenced by other socialization agents that predict future political participation (Levy & Akiva, 2019; Pasek et al., 2008).

We study IPE levels in a school context where the acquisition of self-efficacy, in general, constitutes an important pillar in students’ learning process (Schunk & Paiares, 2002). Self-efficacy, grounded in the larger theoretical framework of social cognitive theory (Bandura, 1977a), affects motivation and behavior. Students feeling efficacious participate “more readily, work harder, persist longer when they encounter difficulties, and achieve at a “higher level” (Schunk & Paiares, 2002, p. 16). This feeling of efficacy develops at a young age and is influenced by the home environment. A stimulating home environment with parents heavily invested in their children’s cognitive development increases feelings of efficacy. Similarly, this is the case for one’s political development, as argued in the first theoretical section. Moreover, stimulating IPE is essential in countering a self-reinforcing spiral between IPE and political participation. After all, when one does not feel capable of participating in politics, one will avoid potential political learning experiences that increase political confidence (Bandura, 1977a). Based on the above, we expect a stimulating political home environment to increase IPE (H1).

Secondly, studies show positive effects of citizenship education on IPE (Pasek et al., 2008; Quintelier, 2009; Quintelier et al., 2012; Westheimer & Kahne, 2004). Moreover, the effect of citizenship education on future political participation is partly mediated by IPE (Levy & Akiva, 2019). This positive citizenship education effect is, however, not straightforward. Studies on self-efficacy show that some school experiences can negatively impact self-efficacy, especially among students less academically prepared to cope with challenging academic tasks (Bandura, 1977b). Not every teaching intervention is necessarily positive toward self-efficacy. For instance, a quasi-experiment of a youth parliament showed a decrease in participants’ IPE (Matthieu et al., 2020), and in a quasi-experiment of on-site citizenship education, there was no effect on IPE (Mulder, 2021). Three citizenship education components are regularly studied internationally: the number of civic learning experiences, the open classroom climate for discussion, and active student participation at school (Schulz et al., 2018). In line with these studies, we assume that more civic learning experiences (H2a), a higher level of open classroom climate for discussion (H2b), and a higher level of active student participation (H2c) increase IPE.

Thirdly, we expect citizenship education to moderate the relationship between a stimulating political home environment and IPE. As argued above, the literature suggests that these effects can be compensating, reproducing, or accelerating. We hope that citizenship education compensates for inequalities in IPE. However, some authors suggest that reproduction and acceleration effects are also plausible. Given these mixed results, the moderating effects of citizenship education may vary depending on the learning strategy. We argue that traditional civic learning opportunities based on knowledge transfers may be more effective in compensating while active learning strategies might be less effective in reducing IPE inequalities due to students’ political home environment. Due to ambiguity about the learning strategy applied for the first citizenship education component, the number of civic learning experiences at school, we solely hypothesize about a moderation effect without specifying the direction. We expect that civic learning opportunities moderate the relationship between a political home environment and IPE (H3a). However, we can study two active citizenship components that might reproduce or even accelerate IPE inequalities: an open classroom climate for discussion and active student participation. Hence, we expect an open classroom climate (H3b) and active student participation (H3c) to reproduce or even accelerate IPE inequalities due to students’ political home environment.

Previous research has found that ability grouping, such as educational tracking, leads to lower self-efficacy among lower-perceived groups (Hoskins et al., 2016; Janmaat et al., 2014; Nieuwelink et al., 2019). As argued above, educational tracking inhibits political learning and reproduces inequalities. Self-efficacy attitudes such as IPE can be expected to be especially sensitive to the stigmatizing effects of educational tracks. We expect students from the technical and vocational tracks to have lower IPE levels than those in the general track (H4). Moreover, the question arises if the moderating effect of citizenship education behaves differently between tracks. As discussed, students from technical and vocational tracks might not only receive less access to qualitative citizenship education (Nieuwelink et al., 2019), but the potential moderation effect of such citizenship education to be a democratic equalizer might also differ between tracks. Students in vocational and technical tracks might not benefit from citizenship education to the same extent because they might have a more challenging time identifying with societies’ view on what kind of person belongs in the political field, namely a highly educated one in correspondence with the diploma democracy perspective (Bovens & Wille, 2017). This can be expected to suppress citizenship education’s potential as a democratic equalizer. We expect that the possible compensating effects of citizenship education components on the relationship between a political home environment and IPE are less strong in the technical and vocational tracks compared to the general track (H5).

Method

Study Setting

We investigate the moderating effects of citizenship education across educational tracks in the 12th grade in Flanders, Belgium’s Dutch-speaking region. Flanders is an often-studied case of a school system that is highly differentiated through early tracking (Vandenbroeck et al., 2022). From the seventh grade onwards, Flanders’s educational system contains educational tracks. Students proceed in two streams in the first two years of secondary education based on prior achievements and perceived abilities. A general academic track with a focus on the general education of subjects such as languages, history, and exact sciences, on the one hand, a (pre)-vocational track providing also general education but at a cognitive less demanding level and focusing more on practical skills. After two years, students proceed with education in four educational forms. General secondary education (ASO) emphasizes a broad general education with a solid foundation for higher education. In technical secondary education (TSO), the focus is mainly on general and technical-theoretical subjects. With a degree in TSO, one can practice a profession or transfer to higher education. Practical lessons are also part of this training. Art secondary education (KSO) links a general, broad education to an active practice of art. With a degree in KSO, one can practice a profession or transfer to higher education. Vocational secondary education (BSO) is a practice-oriented form that focuses on a specific profession and general education (Onderwijs Vlaanderen, n.d.). In Flanders, all tracks share the same citizenship education objectives. These objectives are obtained through a cross-curricular approach, meaning there is no unique citizenship education course in any track. Instead, these objectives are to be pursued throughout the curriculum.

Data

Our data was commissioned by the Flemish educational ministry to test the civic and citizenship educational targets. This data was collected in March 2016 by the support center for test development and polls in collaboration with the department of qualifications and curriculum of AHOVOKS. With a stratified cluster sampling design, they drew a random sample of high schools providing education in the twelfth grade in each educational form. The stratifying variables were educational provider, school type, and urbanization level. In the ASO, TKSO, and BSO samples, 70, 65, and 58 schools participated, constituting a response level of 50, 46, and 41%. These numbers correspond with 1508 ASO students, 1420 TKSO students, and 1185 BSO students from the twelfth grade. 99% of the students in these schools filled out the questionnaire. They also questioned the students’ parents and the school with a corresponding response level of 83% and 90% (Ameel et al., 2016). Simultaneously with this data collection, the International Civic and Citizenship Study (ICCS) took place in the eighth grade, measuring the same constructs. Hence, there is an extensive overlap between the technicalities of the ICCS study and this one (Schulz et al., 2018). We took a subset of the data that only includes ASO, TSO, and BSO students. This means we excluded KSO students since this constitutes a special education with a small n. We omitted classrooms with less than ten students.

Variables

Internal Political Efficacy (IPE)

We measured IPE by relying on the “citizenship self-efficacy” scale developed by the ICCS (Schulz et al., 2016, p. 35). This measurement is closely linked to Bandura’s self-efficacy concept (Bandura, 1977b). The scale includes five items (α = 0.77) measuring students’ self-confidence about doing civic tasks. Including how well they think they would discuss a newspaper article about a conflict between countries; argue their point of view on a controversial political or social issue; run for a school election; follow a televised debate on a contentious issue; and give a presentation in class about a social or political issue. This was measured on a four-point scale ranging from “not well at all” to “very well”.

Political Home (PH)

We consider a stimulating political home environment as a key determinant of IPE. In line with Bourdieu’s conception of cultural capital, we operationalize parents’ capital in the political field as a combination of multiple elements. Firstly, parents’ embodied political capital is measured by parents’ political interest and political talk with their children. On a four-point Likert scale, parents’ political interest was measured by asking the students their perception of their mother’s and father’s interest in political and social issues. The political talk between parents and their children was measured by asking the students how often they talk with their parents about political or social issues, with four answer options ranging from never or rarely to daily or almost daily. Secondly, the objectified cultural capital is measured by asking students how many books they have at home, a standard item used to measure this concept, with five answer options ranging from none to three or more bookcases. Lastly, the parents’ institutionalized cultural capital was measured by asking for the highest educational credential the mother and father obtained. This was asked through the parents’ survey. The political home scale consists of six items (α = 0.70).

Citizenship Education

To study the effect of citizenship education, we include three measurements that tap into different citizenship education components: students’ perceived civic learning experiences (CLE), open classroom climate for discussion (OCC), and active student participation at school (ASP). These measurement scales are developed by the ICCS (Schulz et al., 2016). As stated above, citizenship education is a cross-curricular objective in the Flemish school system. Since there is no separate citizenship education course, we do not have access to an objective measure of how many citizenship education students receive. Therefore, we must rely on self-report measurements capturing students’ perception of received citizenship education. The advantage of this approach is that it allows us to gauge whether moments of citizenship education were effectively transmitted to the students as such. However, we want to acknowledge the limitations of these self-report measures, including potential biases and limitations for generalizability. Due to social desirability, respondents might, for example, overestimate these learning experiences, or recall biases might underestimate them.

Firstly, we test the effects of students’ perceived civic learning experiences. This is a student’s self-report measure gauging how much they have learned about civic education topics at school. This scale includes six items (α = 0.77); how much students have learned about how citizens can vote; how laws are changed; how to solve local problems; how civic rights are protected; political issues and events abroad; and how to look critically at media coverage. This was measured on a four-point scale ranging from “nothing” to “a lot”.

Secondly, the open classroom climate for discussion measures how often teachers encourage students to express their views; share their opinion; talk with others who have different ideas; take the initiative to discuss political topics; express dissenting opinions; and discuss the different sides of an argument while discussing political issues. These six items are measured with a four-point Likert scale ranging from never to often (α = 0.77).

Thirdly, the active student participation scale measures how politically engaged students are in their school. This includes nominating themselves as a class representative or as a member of the student council; voting for class representatives or being involved in the composition of the student council; participating in decision-making about how things are arranged at school; participating in discussions during student meetings; and actively participating in a debate. These five items are measured with three answer categories each: yes, I participated in this in the last 12 months; yes, I participated in this but more than a year ago; no, I have never participated in this before (α = 0.75).

Educational Tracking

A final key variable concerns the educational track of the student. As explained in the study setting, our respondents are categorized into three educational tracks: the academic track ASO, the technical track TSO, and the vocational track BSO. This is a categorical variable with the ASO category as the reference category.Footnote 1

Control Variables

The following control variables are included in the models. Firstly, we included students’ sex with male respondents as the reference category. Secondly, we included students’ immigration status, indicating if the respondent or their parents were born in Belgium as a dummy variable. Respondents with an immigration status are the reference category. These two variables have been shown to affect IPE in previous research (Beaumont, 2011; Gidengil et al., 2008). Thirdly, we included a knowledge test about acting democratically. This test assesses the cross-curricular final objectives regarding citizenship education mentioned above. By including this variable, we isolate differences in IPE while controlling for students’ civic knowledge. The knowledge test items cannot be reported due to legal constraints. All the survey items can be consulted in the online Appendix, and the fit indices of the conducted one-factor confirmatory factor analyses on these measurement scales can be consulted in Supplementary file (see Table S1).

Research Strategy

Before empirical modeling, we standardized the constructed variables. Then, we estimated the intra-class correlation coefficient, revealing that the school level explains a 10,2% variance of IPE. All reported regression models are random intercept models. Since the differences between educational tracks are central to this study, we started by testing the citizenship education access hypothesis. For each citizenship education component, we tested whether students with different political home backgrounds and from different educational tracks indicate different levels of citizenship education. These models are reported in Supplementary file (see Table S2).

We conducted three separate analyses to study the moderating effects of the three citizenship education components across educational tracks. The civic learning experiences models are reported in Supplementary file (see Table S3), the open classroom climate models are reported in Supplementary file (see Table S4), and the active student participation models are reported in Supplementary file (see Table S5). We report three separate models since there are multicollinearity problems when including all the citizenship education components with the three-way interaction terms in one model. All the models are built using a stepwise procedure, the model fits are reported below the tables, and the main models’ generalized variance inflation factors (GVIF) are reported in Supplementary file (see Table S6). The first model of Tables S3–S5 reports the control variables, the second model reports the direct effects of the educational track and political home environment, and the third model reports the direct citizenship education effects. The fourth model includes all the two-way interactions between our key predictors, and the fifth model includes the three-way interactions. Although we only hypothesize the two-way interaction of a political home environment with citizenship education and the three-way interaction to test whether this differs across educational tracks, we also included the un-hypothesized two-way interaction terms in line with Aiken and West (1991) and Song et al. (2019). Not all models have improved fits compared to their previous model, as indicated by the AIC, the BIC, and the (in)significant Chi-square test (see the model fits below the tables). The civic learning experiences models yielded the most reliable results.

Furthermore, to study the interactions more in-depth, we performed additional analyses. Following previous research studying three-way interactions in a multilevel model (Song et al., 2019), we performed group analyses by running the two-way interaction model of a political home and citizenship education on IPE within each educational form in Tables S7–S9 and conducted a simple slope analysis (Aiken & West, 1991). Figures 1, 2, 3, 4, 5, and 6 visualize the significant two-way and three-way regression lines to ease their interpretation. All tables are presented in the appendix, and we report all relevant estimates in the text.

Fig. 1
figure 1

Interaction plot of the marginal effects between internal political efficacy, political home, and the number of civic learning experiences. The plot is made with the R ggpredict function of the ggeffect package, all scale variables were standardized beforehand, and 95% confidence bands are shown. This is plotted from the marginal effects of model 5 of Table S3 in Supplementary file

Full size image
Fig. 2
figure 2

Interaction plot of the marginal effects between internal political efficacy, political home, and the open classroom climate. The plot is made with the R ggpredict function of the ggeffect package, all scale variables were standardized beforehand, and 95% confidence bands are shown. This is plotted from the marginal effects of model 5 of Table S4 in Supplementary file

Full size image
Fig. 3
figure 3

Interaction plot of the marginal effects between internal political efficacy, political home, and the active student participation. The plot is made with the R ggpredict function of the ggeffect package, all scale variables were standardized beforehand, and 95% confidence bands are shown. This is plotted from the marginal effects of model 5 of Table S5 in Supplementary file

Full size image
Fig. 4
figure 4

Interaction plot of the marginal effects between internal political efficacy, political home, civic learning experiences, and educational tracking. The plot is made with the R ggpredict function of the ggeffect package, all scale variables were standardized beforehand, and 95% confidence bands are shown. This is plotted from the marginal effects of model 5 of Table S3 in Supplementary file

Full size image
Fig. 5
figure 5

Interaction plot of the marginal effects between internal political efficacy, political home, open classroom climate, and educational tracking. The plot is made with the R ggpredict function of the ggeffect package, all scale variables were standardized beforehand, and 95% confidence bands are shown. This is plotted from the marginal effects of model 5 of Table S4 in Supplementary file

Full size image
Fig. 6
figure 6

Interaction plot of the marginal effects between internal political efficacy, political home, active student participation, and educational tracking. The plot is made with the R ggpredict function of the ggeffect package, all scale variables were standardized beforehand, and 95% confidence bands are shown. This is plotted from the marginal effects of model 5 of Table S5 in Supplementary file

Full size image

Results

Table S10 in Supplementary file provides a descriptive overview of the key variables used in this study by educational track. It shows that the mean amount of IPE is the highest in the academic track, lower in the technical track, and the lowest in the vocational track (ASO M = 8.8; TSO M = 7.6; BSO M = 6.8, p < 0.001). We observe the same for the political home environment (ASO M = 15.6; TSO M = 12.8; BSO M = 9.3, p < 0.001) and score on the knowledge test (ASO M = 56; TSO M = 49; BSO M = 41, p < 0.001). For citizenship education, students enrolled in the academic track are best off with the highest amounts of reported civic learning experiences, open classroom climate, and active student participation. The same is reflected in Table S2, which tests the citizenship education access hypothesis. These models show how being enrolled in the technical and vocational track is negatively correlated with the number of civic learning experiences (TSO b = -0.23, p < 0.01; BSO b = -0.23, p < 0.01) and active student participation compared to the academic track, the differences are less articulated for the open classroom climate compared to the open classroom climate (TSO b = -0.10, NS; BSO b = -0.13, p < 0.05).

Tables S3–S5 report the models we ran to test the moderating effects of the three citizenship education components across educational tracks. The first hypothesis, expecting a political home environment to increase IPE, is confirmed in the second model (b = 0.17, p < 0.001). We also see significant differences between educational tracks. Students from the technical track indicate significantly lower IPE levels than students from the academic track (b = -0.20, p < 0.001). The same holds for students from the vocational track (b = -0.25, p < 0.001). Hypotheses 2a, 2b, and 2c, assuming a positive effect of citizenship education on IPE, are confirmed in the third model of Tables S3–S5. All three citizenship education components increase IPE (CLE b = 0.19, p < 0.001; OCC b = 0.19, p < 0.001; ASP b = 0.31, p < 0.001).

The fourth model of Tables S3–S5 shows the coefficients of the two-way interaction effects. Hypothesis 3a, assuming civic learning opportunities to moderate the relationship between a political home environment and IPE, is confirmed and shows a compensation effect (CLE b = -0.05, p < 0.01). The marginal effect of this interaction is plotted in Fig. 1. The middle line represents students receiving an average amount of citizenship education, the lowest line represents one standard deviation below the mean, and the highest line represents one standard deviation above the mean. Here, we see a small compensatory effect for civic learning experiences for inequalities in IPE due to students’ political home environment.

Hypotheses 3b and 3c assume that the two active citizenship education components, open classroom climate, and active student participation, might reproduce or even accelerate IPE inequalities. Contrary to this expectation, model 4 of Tables S4–S5 show a compensation effect for both components (OCC b = -0.04, p < 0.05; ASP b = -0.05, p < 0.01). The marginal effects of these interactions are plotted in Figs. 2 and 3 and show also these small compensatory effects of open classroom climate and active student participation for IPE inequalities due to students’ political home environment.

Finally, we tested the direct effects of educational tracking on IPE and whether the moderating effect of citizenship education on the relationship between a political home environment and feelings of IPE is stronger in the academic track than the vocational and technical tracks. We can confirm the fourth hypothesis stating that students from the technical and vocational tracks have lower IPE levels than those in the general track in the second model of Tables S3–S5 (TSO b = -0.20, p < 0.001; BSO b = -0.25, p < 0.001).

The results for hypothesis five are mixed. There is a significant three-way interaction for the civic learning experiences in the technical track (b = 0.10, p < 0.05) and for an open classroom climate in the vocational track (b = 0.09, p < 0.05). The other interaction terms are insignificant. However, the subgroup analyses reported in Supplementary file (Tables S7–S9) indicate larger compensation effects in the academic track compared to the technical and vocational tracks. The subgroup analysis for ASO students indicates clear compensation effects of all citizenship education components (CLE b = -0.10, p < 0.001; OCC b = -0.08, p < 0.01; ASP b = -0.05, p < 0.05). This is not the case for the technical track (CLE b = 0.01, NS; OCC b = -0.02, NS; ASP b = -0.02, NS) and the vocational group has a significant small compensation effect for civic learning experiences and active student participation (CLE b = -0.07, p < 0.05; OCC b = 0.01, NS; ASP b = -0.08, p < 0.05). The marginal effects between IPE, political home environment, and the three citizenship education components across educational tracks are plotted in Figs. 4, 5, and 6. These plots also show clearer compensation effects for the academic track than the other tracks. Although we see a similar pattern in the vocational track for civic learning experiences and active student participation, there is a high overlap of error bounds. Hence, we cannot discern a clear compensation effect in the vocational track. With this evidence, we cautiously confirm the fifth hypothesis that assumes a stronger compensation effect in the academic track than the technical and vocational track and might indicate a reproduction of pre-existing inequalities in IPE.

Discussion and Conclusion

This study investigates whether citizenship education compensates, reproduces, or accelerates inequalities in IPE across educational tracks. Firstly, our analysis shows that IPE is unequally distributed. Youngsters growing up with parents having higher amounts of capital in the political field – meaning their parents show an interest and talk about politics with their children (embodied political capital), have many books at home (objectified cultural capital), and have high educational credentials (institutionalized cultural capital) – show higher amounts of IPE. This indicates that parents’ political habitus provides their offspring with political resources and experiences, resulting in children feeling capable of acting in the political field. This finding recognizes parents as important socialization agents in their children’s political development (Neundorf & Smets, 2017) while controlling for students’ civic knowledge. Hence, inequalities in IPE cannot be reduced to skill-based explanations. The home environment, at least partly, shapes students’ feelings of ability in the political field.

These inequalities undermine the democratic principle of equal voice and opportunity (Dahl, 2006). Regardless of their home situation, all citizens deserve the necessary resources to take up their democratic role. Herein, schools are believed to play a pivotal role. They hope to achieve a compensation effect by providing citizenship education. Our analysis shows how citizenship education might be beneficial to increase IPE and compensate for a lack in a political home environment. We find both direct positive effects of all three citizenship education components and modest yet significant compensation effects. However, we also show how access to this citizenship education is unequally distributed, with academic track students indicating higher access to citizenship education, especially civic learning experiences and active student participation, besides their higher initial levels of IPE. Furthermore, the compensation effects of citizenship education might also differ across educational tracks. Our results cautiously indicate how citizenship education might be more successful in working as a democratic equalizer in the academic track than the technical and vocational tracks.

This might reproduce pre-existing inequalities in IPE (Hoskins et al., 2016; Janmaat et al., 2014). Students in vocational tracks experience less encouragement to become politically active than students in academic tracks (Nieuwelink et al., 2019), which is also reflected in our results. Citizenship education succeeds in creating more equal levels of IPE, but this is not as articulated in the technical and vocational tracks. Besides unequal access to this citizenship education, the quality of this citizenship education might also differ. Nieuwelink et al.’s (2019) qualitative study describes this by showing how students in academic tracks have better opportunities to discuss politics in class and are encouraged to become politically engaged, exacerbating instead of compensating inequalities. Future research should consider this and explain how inequalities in political behavior are reproduced in a school context and how to counter it.

Our results should be carefully interpreted within the context of cross-sectional data in a single case. We cannot claim strong causal inferences for the compensatory effects we find. Future longitudinal or experimental studies could solidify these relations. Nonetheless, our results align with other qualitative and quantitative studies and support the argument that it is possible to compensate for the lack of a political home environment with citizenship education, but this might depend on the educational track. Although we could not distinguish different moderation effects based on citizenship education type, it is worth exploring if different learning strategies function as successfully as democratic equalizers. Moreover, while our results from the Flemish case are likely to generalize to similar European countries that implement early tracking such as Germany, the Netherlands, and Austria, they are unlikely to generalize to contexts with late or no institutionalized tracking.

It is a hopeful finding that citizenship education seems to compensate for the lack of a political home in the academic track. However, our results warrant further worries about the diploma democracy that characterizes our political system (Bovens & Wille, 2017). The results of this study suggest that policymakers not only consider equal access to citizenship education, especially for technical and vocational students, but also reflect upon the socialization of class-based identities regarding political engagement during citizenship education.