UvA-DARE (Digital Academic Repository) What do adolescents know about citizenship? Measuring student’s knowledge of the social and political aspects of citizenship

: This paper analyses how young people’s citizenship knowledge is related to the di ﬀ erent domains of citizenship in their daily lives. Based on a representative sample of some 5300 students in the third year of 80 Dutch secondary schools, our study relates citizenship knowledge to student background and school characteristics. The knowledge test developed for this study situates citizenship knowledge in the literature and the societal and political context deﬁning the social structure students live in. The contribution of our study lies in this broad conceptualisation of citizenship, which is reﬂected in ﬁne-grained, more speciﬁc results than the outcomes of earlier research. Gender di ﬀ erences are particularly pronounced in the social aspects of citizenship and are small in the political domain. As far as ethnic background is concerned, we see knowledge di ﬀ erences in the domain of “acting democratically”. This is also the domain where most of the di ﬀ erences in citizenship knowledge between students of the various schools and tracks occur. School size, public / private school, urbanisation and a more heterogeneous student population cannot explain these di ﬀ erences. To mitigate inequalities in citizenship knowledge between and within schools, which are relatively large in the Netherlands, further research is necessary to investigate micro-level mechanisms within schools.


Introduction
A strong and resilient democracy needs citizens who have the competences to participate in various forms of social and political life (Barber 1984).In this respect, adolescence is an important phase as young people are relatively open to changes in their views on a democratic way of living that have a lasting impact (e.g., Russo and Stattin 2017;Shehata and Amnå 2017).Hence, citizenship and the promotion of citizenship competences through education have long been an important issue in politics, the public debate and science.This applies to many countries (e.g., Eurydice 2017; Haste 2010).Knowledge is especially important for young people's citizenship competences.Support for democracy, for example, is partly dependent on knowledge of democratic values (Cho 2014).Various researchers also point out that political knowledge is likely to have a positive influence on political interest and participation (Dassonneville et al. 2012).In addition, empirical research shows that schools especially seem to influence the knowledge component of citizenship competences (Coopmans et al. 2020;Isac et al. 2014;Schulz et al. 2018).
Citizenship knowledge is not equally divided among adolescent groups (Schulz et al. 2018).Social background factors, such as gender, ethnicity and socio-economic status (SES) of the family, in particular, play a role.The differences are relatively large in the Netherlands, where this study was carried out.Not only do Dutch students have less citizenship knowledge on average than their peers in comparable countries participating in ICCS2016 (Belgium/Flanders), Denmark, Finland, Norway and Sweden), the variance between the various knowledge Levels is also considerable (Dijkstra et al. 2021;Munniksma et al. 2017).The strong external differentiation in the national education system in the form of separate school types is reflected in the citizenship results.Students in higher school types have more citizenship knowledge and the difference with students in lower school types is growing between 2009 and 2016.
Typically, standardized multiple-choice knowledge tests are the main instrument for investigating adolescent citizenship knowledge.Examples include the international ICCS study, referred to above, that is being periodically conducted (Schulz et al. 2018) and the CELS cohort study in England (Keating et al. 2010).These knowledge tests are necessarily limited in scope because they are part of a measurement that includes attitudes, skills and political participation.Moreover, the ICCS study has operationalized citizenship knowledge in a way that allows comparison of results across countries.Obviously, this limits the possibilities to ground citizenship knowledge in the political and social context of specific countries.As early as the 1960s, Almond and Verba (1963) pointed out the relevance of a country's political culture, which transcends the formal institutions of democracy.The democratic and constitutional values of citizens who support democracy as a system of governance are related to current practices that make up peoples' lives.Later research indeed showed that the daily democratic experiences of citizens and their perceptions of how democracy works influence their opinions (Alonso 2016;Biesta et al. 2009;Nieuwelink et al. 2018).Factors such as a country's culture (Barrett and Brunton-Smith 2014), income distribution (Andersen 2012), migration history and the integration of minorities (Heath et al. 2014) also influence citizens views and values.Finally, statutory regulations concerning citizenship education, the education system and the schools' teaching practices vary considerably across countries (Eurydice 2017) and thereby affect democratic attitudes of adolescents and their citizenship knowledge.
In this article, we approach citizenship knowledge from a broader perspective than we find in the scientific literature to date.First, we emphasize the social aspects of citizenship more strongly than has been done so far in research.Secondly, we situate citizenship knowledge not only in the literature on citizenship but also in the societal and political context defining the social context students live in.Based on such a broad conceptualisation and measurement of citizenship, we use a recent, extensive knowledge test, thoroughly investigating the citizenship knowledge of Dutch adolescents.Our aim is to provide schools and policymakers with tools to strengthen student citizenship knowledge and to help them reduce the knowledge gap between different groups of adolescents.
Our research question reads as follows: To what extent does the citizenship knowledge of Dutch adolescents relate to individual background characteristics (gender, ethnicity and SES) and characteristics of the schools teaching these students (school size type, public/private school and degree of urbanisation), and to what extent do differences in knowledge relate to the different domains of citizenship in the adolescents' daily lives?In the next section, we first discuss the concept of citizenship and the relevance of citizenship knowledge.We then outline how we approached citizenship knowledge in the societal and political context of Dutch students.After describing the method and the empirical results of our study, we summarize our main findings and reflect upon them by comparing them with the results of earlier research, especially the ICCS 2016 study, in which the Netherlands participated in the same period.

Citizenship and the Importance of Citizenship Knowledge
In general, the common interpretation of citizenship in the literature on citizenship education concerns the competences needed to participate in by definition "hybrid" cultures (Eidhof et al. 2016;Haste 2010;Knight Abowitz and Harnish 2006).Citizenship refers to the competences that people need to participate in a pluralistic and multicultural society, to function in the democratic constitutional state and to maintain it.Citizenship in societies characterized by diversity and relevant to all students not only includes the "minor virtues" (such as social competences, good manners, etc.) but also being able to deal with differences and conflicts, being willing and able to critically evaluate different perspectives, exploring strategies for change, reflecting on justice, inequality and moral action and having a democratic attitude and acting in a democratic manner (Geijsel et al. 2012;Westheimer and Kahne 2004).In a strong democracy, the participation of all people, from all groups in society, in all forms of societal and political life is required (Barber 1984).Consequently, citizenship refers to the agency of individuals not only within the domain of politics but also within the wider social domain and citizen relationships outside the domain of governance.
The development of citizenship competences cannot be taken for granted.Among other socializing agents, the role of schools is particularly important because they reach most adolescents.Fostering "educated citizenship" is therefore considered as one of the main tasks of schools (Campbell et al. 2012).Citizenship knowledge of students is an important pillar of citizenship education.This concerns both factual knowledge, concerning for example fundamentals such as terms and democratic principles, as well as conceptual knowledge about the relationships between these elements and procedural knowledge that refers to how to act responsibly (Geijsel et al. 2012;Schulz et al. 2018).Research shows that citizenship knowledge is related to citizenship skills, attitudes and behaviour.There is a positive link, for example, between citizenship knowledge and self-reported citizenship competence (e.g., Manganelli et al. 2015;Ten Dam et al. 2011) and between citizenship knowledge and citizenship attitudes (e.g., Isac et al. 2014;Lauglo 2013;Ten Dam et al. 2011).Fourteen-year-old students with more knowledge of citizenship appear to be stronger advocates of ethnic and gender equality and have a greater interest in politics (Lauglo 2013).Citizenship knowledge and behaviour are also related.Students with more citizenship knowledge are more likely to say that they will vote once they are entitled to do so and indeed they do (e.g., Castillo et al. 2015;Cohen and Chaffee 2013).These relationships can be understood, inter alia, on the basis of the informative function of citizenship knowledge (Galston 2001), which enables citizens to form an opinion on citizenship issues and to act when necessary.Knowledge of citizenship also makes people aware of other perspectives and can thus encourage them to think and act in the public interest.Moreover, citizenship knowledge contributes to their identification with social themes and promotes social trust and engagement (Van Elsas 2015;Van Ingen and Bekkers 2015).Reichert (2016) adds that citizenship knowledge enhances self-confidence in the social domain and feeling informed and able to act; citizenship knowledge thus fosters self-efficacy, which, in turn, contributes to real action.
Questions of conceptualisation (how to define citizenship) play an important role in determining student knowledge levels.If citizenship is not just something for "later in life" but also matters to the students' lives now, their world and how they live in it is relevant too.The instruments available to date differ about this aspect.The Citizenship Education Longitudinal Study (CELS; Keating et al. 2010) and the National Assessment of Educational Progress (NAEP; U.S. Department of Education, Institute of Education Sciences, National Center for Education Statistics 2010) used in the US focus on knowledge of national politics and constitutional democracy.To some extent, this also applies to ICCS 2009.These studies mainly focus on citizenship as the relationship between the individual and the democratic constitutional state.Moreover, they regard young people primarily as future citizens.The knowledge tool of ICCS 2016 includes a broader range of knowledge and also focuses on citizenship issues within a societal context such as dealing with conflicts, sustainability, social media and cultural differences.
In addition to the political dimension of citizenship, this places more emphasis on the social component and addresses everyday citizenship in actual practice, including societies' opportunities structure to a greater extent.We consider this to be crucial for a socially relevant and valid measurement of young people's citizenship.

Measuring a Broad Concept of Citizenship Knowledge in the Dutch Context
The increasing importance of citizenship education in the Netherlands, as in many countries, and the limited availability of instruments for measuring citizenship knowledge that can be used by schools prompted the development of an instrument for measuring citizenship knowledge, in which citizenship includes aspects of both the social dimension (e.g., knowledge of dealing with conflict, critically evaluate different perspectives, strategies for change, social behavior, norms and values, lifestyle, environment) and the political dimension (knowledge of the democratic constitutional state and democratic principles).We took four social tasks which are representative of and meaningful for the actual citizenship practices of young people as a starting point (see Table 1) (Ten Dam and Volman 2007;Ten Dam et al. 2011).
The social task "acting democratically" concerns knowledge of the democratic constitutional state and its underlying basic values and principles and knowing what it means to act in accordance with those values and principles.The democratic constitutional state emphasises the role of citizens (principle of equality and the citizen's right to influence and monitor government) and the trias politica.Additionally, the main political movements (like liberal, socialist or general communitarianism), the main features of the European Union (economic cooperation, promoting security and prosperity) and the role of citizens within the EU were taken into account.The basic set of commonly held democratic values and principles consists of liberty, equality (including tolerance and non-discrimination) and solidarity (Eidhof et al. 2016;Van Goethem et al.).These are, in one way or another, enshrined in many national constitutions, as the Dutch Constitution and the universal declarations on human and children's rights (UN General Assembly 1948, 1959).As part of the social task "acting democratically", special attention should also be paid to democratic decision-making and democratic participation.Democratic decision-making involves knowledge of the characteristics (e.g., enabling participation for all and including dissenting and minority views) and relevance (e.g., protecting and representing the public interest) of inclusive democratic decision-making (Nieuwelink et al. 2016).Democratic participation concerns empowering and involving citizens in the democratic constitutional state and its governance.Well-considered opinions require knowledge of how to use sources of information responsibly and an understanding of the influence of one's frame of reference (Tully and Vraga 2017).It is also important to have knowledge of the various ways of exercising control and participation (e.g., social and other media, voting, petitioning, going on strike) and to be able to assess their democratic value (Hooghe et al. 2016;Segal et al. 2017).The democratic principle of liberty and freedom of speech plays an important role here.
The social task "acting in a socially responsible manner" concerns knowledge about the interdependency of citizens and government in the domains of the economy, mental and physical health and the environment (Ainley et al. 2013;Oxley and Morris 2013;Westheimer and Kahne 2004).Within this context, we also aimed to pay explicit attention to self-reliance (Kinnier et al. 2000), social conventions and social responsibilities, and the rules and mores that strengthen social relations and promote public order such as people's conduct in traffic, at school, at work and in the public space (Van Goethem et al.;Smetana et al. 2014).
The social task "dealing with conflicts" represents how people respond to discord, for example, in the form of incompatible or wayward behaviour or opinions (Laursen and Hafen 2010).Peaceful conflict resolution is widely seen as a crucial aspect of citizenship (Eidhof et al. 2016).The knowledge element relates primarily to causes of conflict, which involves respect or disrespect for democratic principles and values.Inequality or disagreement about property or values, for example, and the inability to meet basic needs such as respect and security are major causes of conflict (Fisher et al. 2013).Knowledge of modes of conflict resolution is also important, for example, ways of taking as much responsibility as possible and taking into account and making room for people's perspectives and their interests (Marceau et al. 2015) and knowing why and to what end specific ways of handling conflicts are constructive (e.g., win-win situation and restoring relationships; see Fisher et al. 2013) or not or less productive (e.g., avoiding all conflicts, focusing on negative emotions and conflict escalation; see Marceau et al. 2015).

Dealing with differences
Meaning and knowledge of differences and ways of dealing with these differences at social level

Meaning and knowledge of differences
Meaning of differences at social level Exemplary knowledge of social differences Dealing with differences Adaptive and maladaptive ways and results of dealing with differences Finally, the social task "dealing with differences" is increasingly important in a plural, multicultural and global society.This may concern socioeconomic, social, religious, cultural, lifestyle and external/physical differences.In order to handle differences well, it is important to know differences and ways in which differences can influence behaviour (Banks 2001;Kymlicka 2003).Democratic values such as equality, tolerance and the principle of individual autonomy play an important role in dealing constructively with differences, for example by respecting and tolerating differences, applying universal values and looking for shared values and interests (Eidhof et al. 2016).Strategies such as imposing one's norms and values, isolation and exclusion, polarisation and discrimination are in stark contrast to the above (Putnam 2007).
The conceptual framework that has been briefly described above is summarised in Table 1.It formed the basis for the development of our instrument.
Our study took place in the Netherlands.Promoting citizenship competences is a statutory task for Dutch schools and the objectives of both primary and secondary education are stipulated in legislation since 2006.Due to the constitutional "freedom of education" schools have a high degree of autonomy.They can design citizenship education as they see fit, as long as it respects the basic values of the democratic state and is based on an underlying vision aimed at active citizenship and social integration.How schools meet the citizenship obligation (e.g., teaching it as a separate subject, or through projects or embedded in other subjects), what content they teach and how much attention they pay to it is up to each school individually.A final important characteristic is the high degree of external differentiation in the Dutch education system (Van de Werfhorst 2014), with students being placed in various tracks at an early age (12 years) in prevocational and general education.

Construction of Citizenship Knowledge Instrument: Content
First, we developed a conceptual framework on the basis of about 40 review articles on citizenship.Data files in the Social Sciences Citation Index (SSCI, accessible via Internet) were checked for potentially relevant studies published between 2000-2017, using the following keywords, including synonyms and closely related terms: citizenship education, moral education, character education, intercultural education, values education, democratic education, education for life, civic competence, and civic education, social competence, personal and social education, and affective competence.We restricted our selection to articles published in academic, refereed periodicals, papers presented at international congresses using a peer-review selection procedure, books and contributions to books.We operationalised young people's citizenship knowledge into four social tasks which are considered exemplary for their actual citizenship practice: acting democratically, acting in a social responsible manner, dealing with conflicts and dealing with differences (Ten Dam and Volman 2007;Ten Dam et al. 2011; see Theoretical framework).In addition, core elements of citizenship were identified on the basis of the universal children's and human rights treaties (UN Charter and Universal Declaration of Human Rights).To do justice to the national context, we further looked into the statutory citizenship obligation and the underlying Dutch constitution, the associated assessment framework of the National Inspectorate of Education and the attainment targets for civics in Dutch secondary education (Inspectie van het Onderwijs 2006).
Secondly, we supplemented the identified scientific and legal notions of citizenship knowledge with societal views about what citizenship knowledge is important for students leaving secondary education.From an educational perspective, opinions in society on the civic knowledge that young people should acquire are not only relevant for the conceptualisation of the concept of citizenship as such or its relevance to curriculum design.Operationalisations that also take account of societal views on desirable civic knowledge are important because of the potentially large impact of outcome measurements in the domain of civic knowledge, for example for research into school effectiveness and as a basis for monitoring with a view to education policies and school improvement.We therefore conducted a mixed-method study of the views in society about what citizenship knowledge young people should have and thus should be taught at school (Van Goethem et al.).We examined questionnaire data from a representative national sample and supplemented this with data from a multi-case study among several focus groups of stakeholders with various roles and interests in Dutch society (e.g., trade unions, administrators, employers, teachers, students and charities).The topics considered most important were values and norms, rules of conduct, human rights (including children's rights), a responsible and healthy lifestyle and knowledge about what democracy entails.
Based on the above two steps, we developed an instrument that can be used in quantitative, large-scale research in which groups of students and schools can be compared with each other.In order to also be able to benchmark the results of our study with those from other studies, we used comparable student and school characteristics.We developed a multiple-choice test with three response options for each question (dichotomous measurement).

Construction of Citizenship Knowledge Instrument: Design and Analyses
We decided on a design with anchor items.All students were presented with the same set of 26 items (the anchor items), high-quality items from the pilot study for all domains (12 items on acting democratically, 4 on acting in a socially responsible manner, 4 on dealing with conflicts and 6 on dealing with differences).Fourteen test versions of 40 items each were used.Each version contained the 26 anchor items and 14 items that were only included in one set.The total set thus comprised 222 items, 196 (= 14 × 14) of which were distributed across the 14 test versions.The other 26 items were presented to all respondents.This design led to each non-anchor item being completed by a minimum of 300 respondents, a suitable number for estimating item and respondent parameters as part of an IRT analysis (De Ayala 2009).
Both the pilot and the main study were preceded by an item analysis: various aspects of the quality of all items were examined.

1.
The difficulty of the items was examined by looking at the proportion of correct answers per item.
The reliability of a test is highest when the difference in the degree of difficulty between items is small and the degree of item difficulty is a little less than halfway between the chance level and 1.00 (Lord 1952).In this case with three answer options, the ideal value is just under 0.67 (= 0.33 + (1 − 0.33)/2).As a rule of thumb, the degree of difficulty here may be two-tenths under or over, and thus in this case should preferably be between 0.47 and 0.87 (Meyer 2014).

2.
An item total correlation was calculated per item to gain an indication of the item discrimination.
Item discrimination is the extent to which an item can differentiate between students with different scores in the test.Low discrimination, for instance, means that an item only differentiates between students with very different scores in the test.As a rule of thumb, item discrimination values should be somewhere between 0.3 and 0.7 (Meyer 2014).

3.
Next, a distractor analysis was carried out.We looked at whether the correct answer was chosen most often by the respondents and whether the incorrect answers (distractors) were not selected by too few students (for this we set a lower limit of 4%).Because, if an incorrect answer (distractor) is rarely selected, the alternative was too clearly incorrect, meaning that in essence the question had two rather than three alternative answers.We also looked at the item-total correlations of the distractors, which, ideally speaking, should be negative.For instance, if there are positive 1 Prior to participation, and in line with the General Data Protection Regulation at the time, letters were sent to students' parents, explaining the study aims and procedures.With this letter, parents were given the opportunity to deny permission for their child's participation.
item-total correlations with the correct answer, but also with an incorrect (distractor) answer, then there is no clearly correct answer to the question.4.
We used the program jMetrik 4.0.5 (Meyer 2014) to carry out DIF analyses for gender, maternal ethnicity, SES and school type, checking whether the items were valid for these different groups (e.g., Mellenbergh 2005).This was assessed by making a comparison with the total competence of the students.For example, if female students completed a certain item systematically better than expected on the grounds of their total competence, then for that item there was a case of "bias" towards female students, and thus differential item functioning.5.
Finally, we used Confirmatory Factor Analyses (CFA) to test whether the theoretic distribution of the items in the four social tasks was also found in the data, checking whether the models fitted sufficiently (RMSEA < 0.05; CFI > 0.95; TLI > 0.95; WRMR < 1.00; Bollen and Long 1993).
The results were used to examine the formulation of the items, after which the formulations in part of the items in the pilot were refined where possible.The items in the main study that were satisfactory on the basis of these analyses were included in the instrument.Following this, reliability estimates were made for each of the four social tasks.Because not all students completed all items, the following procedure was used: for each task, an IRT scale (item-response theory) was created using the program OPLM (Verhelst et al. 1995).For every item in the test versions, a difficulty was determined that can be compared across the versions.Because the distribution of respondent parameters can also be estimated, the reliability of the scale can be determined for the total sample of students, but also broken down into samples per school type.To measure the degree of reliability, we used the MAcc index from OPLM.This reliability measure is comparable to the coefficient Alpha (Bechger et al. 2003;Verhelst et al. 1995, pp. 99-100).

Respondents and Procedure
A stratified sample was taken based on the distribution of the following school types: 1.
general secondary education or pre-university education 3. mixed The final selection of schools (n = 82) is representative in terms of school type, geographical distribution, school denomination (public, private religious, private non-religious), degree of urbanisation and school size.
In the spring of 2016, the citizenship knowledge test was completed by 5301 students in the third year of all schools.To determine ethnicity, the migration background of students was used as a proxy.In accordance with the Statistics Netherlands (CBS) definition, a migration background was assumed if at least one of the parents was not born in the Netherlands. 2Table 2 presents the sample data.In each school, the 14 sets of 40 knowledge items each were randomly assigned to the students.Each set was completed by a minimum of 350 students.Completing the test took no more than 30 min.It was digitally and anonymously administered by trained test leaders (the results were not shared with teachers, peers and parents).The test was administered as part of a study of the role of the school in the development of the citizenship competences of secondary school students.

2
A limited proportion of students with a migration background was not born in the Netherlands (first generation): for Turkish 8%, Moroccan 5.1% and Surinam/Antilles/Aruba 8.7%.In order to find out to what extent students with a migration background of the third generation were born in the same country as their father or mother, a separate question about country of birth was asked about all four grandparents.This approach shows that students with a migration background are largely of the second generation in this study.Because there are too few first and third generation students in the study, making a distinction between generations in the analyses is not possible.

Analysis
Variance analysis and post hoc comparisons (Bonferroni for equal variances and Dunnet's T for unequal variances) allowed comparison of the scores and standard deviations for the total citizenship knowledge scores and the scores on each social task of the various student groups.Because the group variables are interrelated, generalized linear model analyses were conducted out to control for this.
Table 3 presents the number of items for each scale for the four social tasks, the average scale scores and standard deviations, the correlations between the social tasks and the reliability coefficients for the scale scores.The social tasks appear to be separate factors: there is a mean correlation between the levels of students for citizenship knowledge in the four social tasks (range: r = 0.53 ** to r = 0.60 **).It also appears that reliable scale scores can be formed for total citizenship knowledge and for the four social tasks.When looking at each of the tasks individually and the total population of students, the reliability of the scale based on all available items is between 0.855 and 0.938.In the individual school types, we see the highest reliability for all tasks for lower and higher secondary education.For the other school types, the reliability was always higher than 0.7, except for "acting in a socially responsible manner" at pre-university education, where reliability was 0.667.However, at the gymnasium (the highest of pre-university education) schools, the reliability for this task was 0.763, so this should not be seen as an indication of a ceiling effect.
Table 3. Descriptive details of the citizenship knowledge instrument per social task, and the total instrument and correlations between the social tasks.The instrument meets the common requirements of validity and reliability (see Appendix A).

The Influence of Student and School Characteristics
To answer the question of how young people's citizenship knowledge is related to, on one hand, individual background characteristics and, on the other hand, school characteristics, we estimated multivariate multilevel models with citizenship knowledge related to the four social tasks as the dependent variables, at the level of students and schools.The models added the student characteristics first and then the school characteristics.Student characteristics include the parental level of education (highest level reported, including "unknown"), the students' ethnic origin (including "unknown") and the educational level (school type) at which the students are being taught.The school characteristics that were successively modelled were school size, public/private school (dichotomy: public, yes/no), degree of urbanisation (1, non-urban to 5, metropolitan), social composition (percentage of students with parents with a low level of education), the ethnic composition of the school (Herfindahl index, based on the percentages of students of a particular ethnic group) and the presence of other school types within the school in addition to the student's (single track school: yes/no).
Based on the 0-model, the variance at school level (ICC) differed for the various domains of citizenship: acting democratically (27%), acting in a socially responsible manner (16%), dealing with conflicts (17%) and dealing with differences (20%).
The first series of models (see Table 4A-D) focused on estimating the relationship between citizenship knowledge and student background characteristics.Generally speaking, this is the case for gender, parental level of education and student ethnic background: female students generally know more than male students, students from families with a higher parental level of education score higher on the knowledge test than other students and the knowledge level of students with a migrant background is on average lower than that of native Dutch students.The latter, however, does not apply to students from the former Dutch colonies (Surinam, the Dutch Antilles, Aruba).The knowledge level of students with a western migrant background does not differ from that of native students without a migrant background either.In addition, we see differences between students in different school types: the level of knowledge of students in higher tracks is higher than that of students in lower tracks.
Although school type is included in the models as an individual student characteristic, students often attend classes with peers in the same or adjacent tracks, which means that school type has, in fact, a dual nature: apart from being an individual proxy of student performance, it is also a grouping characteristic at school level.This is reflected by the analyses: the model in which school type is added with some other characteristics did cause an increase in explained variance at school level of between 65.4% and 79.1% (see Table 4A-D).
The student knowledge level did not appear to be related to school size, the public or private character of the school or degree of urbanisation.Nor did we see a difference in knowledge level associated with a more heterogeneous student population in terms of parental level of education.

The Influence of Student and School Characteristics on the Various Citizenship Domains
In our study, we also investigated the relationship between differences in citizenship knowledge and the various domains of citizenship in daily life.Table 4E presents the results of the last model (3) on the four social tasks.This concerns the differences in student knowledge after controlling for all other characteristics in the model.We present effect sizes to show how large these differences are and to what extent there are differences in effects between the social tasks.
For the four social tasks, the gender differences (with female students scoring higher) appear to be greatest in the domains of acting in a socially responsible manner and dealing with conflicts (small to medium effect sizes).The gender differences are smallest in the domain of acting democratically.The differences found for parental level of education are almost negligible; the differences that do occur were found in the domains of acting democratically and dealing with differences.The knowledge difference between students with and without a migrant background was small.We saw it particularly in the domain of acting democratically (small to medium effect sizes).Students of Moroccan descent scored lower on the knowledge test than native Dutch students (non-migrant background) in all areas of citizenship.
For all four citizenship domains, we saw medium or large school differences.These differences in knowledge level increase with track level.This track effect was the greatest in the domain of acting democratically.The school's ethnic diversity only played a role in the domains of acting in a socially responsible manner and dealing with conflicts.The effect size was nearly medium: the greater the diversity of the school's student population, the lower its average knowledge level.As stated, we did not find a relationship between student citizenship knowledge and the other school variables (school size, public/private school, degree of urbanisation, the percentage of students with parents with a low level of education, single or multi-track school).

Conclusions and Discussion
This article describes a detailed, representative analysis of the citizenship knowledge of Dutch students in grade 9, relating it to both student background and school characteristics.In our study, citizenship is considered to be situated in the daily social practices of young people and operationalised in terms of the knowledge that they need to adequately fulfil four social tasks: acting democratically, acting in a socially responsible manner, dealing with conflicts and dealing with differences.The knowledge test covers a wide variety of citizenship aspects, rooted in the literature and the societal and political context defining the social structure students live in, as well as international human rights treaties, the national constitution, the statutory citizenship obligation and the associated framework of the Inspectorate of Education and the attainment targets in Dutch secondary education.
The instrument builds on the-still young-tradition of evaluating students' citizenship knowledge through standardized measurement, focusing on conceptualisation of the relevance of citizenship within the social context of young people, and especially on its underlying societal and political opportunity structures.These structures not only determine the (unequal) opportunities for political participation, but also determine the chance of success in later life.The socio-economic and social inequality in modern society shape the different experiences of young people and, as a result, their potential gains in participating in democracy, economy and society at large.Modern societies are characterized not only by inequality in the chances of successful participation in society, but also by a high degree of diversity.While the promotion of citizenship knowledge for all students could contribute to an inclusive society with equal opportunities for social, cultural and economic participation, the far-reaching social inequalities also mean that the benefits of acquiring citizenship knowledge vary between groups.As a result, optimal equipment for participation for all in societies characterized by diversity as well as inequality, asks for citizenship knowledge concerning difference, conflict, divergent perspectives, social change and the importance of critical capacities.Focusing on four social tasks in which these aspects are important features and linked to citizenship that is relevant in the daily context of young people, our conceptual frame maps citizenship knowledge in such a way that justice is done to dealing with diverse perspectives, critical reflection and the ability to change, as well as the values for perpetuating an open, democratic society, of which equality, tolerance and freedom are the constituent characteristics.This combined focus on peaceful coexistence and attention to diversity and difference provides a relevant starting point for describing and reflecting on the acquisition of citizenship knowledge through education.
That, as we will discuss further below, the measurements show differences between students from different ethnic groups and with high and low educated parents, is therefore an important result.Broadly speaking, the parental educational level does not correlate with different civic knowledge outcomes, as is the case for ethnic background.Of the diverse ethnic groups, however, Moroccan students stand out, with low scores on all components.Unlike other minority groups, the lagging position of this group on other educational outcomes, such as dropout or level of achievement, is also visible in the social domain.This finding, on one hand, underlines the importance of our conceptualisation of citizenship and, on the other, it makes it clear that the acquirement of citizenship knowledge cannot be separated from the social position of groups in society.
The results of our study show both similarities and differences with earlier cross-sectional research.First, it appears once again that differences in citizenship knowledge are largely attributable to student characteristics.The variance between schools is smaller and is explained for a substantial part by their student population (Dijkstra et al. 2015;Isac et al. 2011;Isac et al. 2014;Schulz et al. 2010Schulz et al. , 2018)).When it comes to differences between students, the effects found were largely in the same direction as those found in other studies.However, the broad conceptualisation of citizenship used in this study is reflected in its more specific and fine-grained results.
In line with earlier studies on citizenship knowledge among Dutch students (Geijsel et al. 2012), female adolescents have more citizenship knowledge than male students.ICCS 2016 shows the same, although gender differences are smaller in the Netherlands than in countries with similar positions on the Human Development Index of the United Nations Development Programme (Dijkstra et al. 2021;Munniksma et al. 2017).A more detailed breakdown by domains of citizenship reveals that gender differences mainly occur in the social aspects of citizenship (acting in a socially responsible manner and dealing with conflicts).When it comes to the more political aspects of citizenship (acting democratically), the differences between female and male students are smaller.These are precisely the aspects of citizenship that prevail in the ICCS study 2016, which distinguished between Citizenship society and systems (society and its institutions, organisations and systems), Citizenship principles (democratic principles such as equality, liberty and the constitutional state), Citizenship participation (forms of citizenship participation) and Citizenship identities (self-identity and links with society).The ICCS 2009 study focused on the political domain of citizenship even more.Its results show that female students know more about citizenship than male students except in the Netherlands (Schulz et al. 2010).It is not surprising that the way in which citizenship knowledge is operationalized has an impact on the results.However, it is also possible that the Dutch curriculum, which gives schools much leeway to set their own goals, the place of citizenship teaching in higher grades, may explain the different pattern of citizenship knowledge among Dutch female and male adolescents.It will take a more in-depth analysis of the relative influence of the curriculum on the various aspects of young people's citizenship knowledge to obtain a better understanding of this issue.
Further, in line with earlier research (Geijsel et al. 2012;Isac et al. 2011;Schulz et al. 2018), we find a relatively high knowledge level among students with better-educated parents.These differences are small, however.The fact that the knowledge differences are small is probably due to the Dutch education system (see below).If we zoom in on the different domains of citizenship, we notice that parental level of education does not have an impact at all on knowledge of the social task of dealing with conflicts.
In line with the ICCS findings (Munniksma et al. 2017;Schulz et al. 2010Schulz et al. , 2018)), this study also showed that a migrant background has a negative impact on student knowledge.However, if we specifically look at the specific national situation, it appears that the knowledge level of students originating from the former Dutch colonies (Surinam, the Dutch Antilles, Aruba) is similar to that of native Dutch students.This applies to all domains of citizenship.This student group has not necessarily lived longer in the Netherlands than students of Turkish or Moroccan descent (the most important migrant groups in the Netherlands).A possible explanation is that the school curriculum in the former colonies was based on the Dutch curriculum for a long time and still bears traces of it.Of the migrant groups in the Netherlands, the students of Moroccan descent have a lower level of knowledge of all four social tasks, and especially acting democratically, than the Dutch students without a migrant background.Their relatively low level of citizenship knowledge may be related to their response to the negative image of migrants in the Netherlands.Research indicates that adolescents of Moroccan descent "suffer more than their Turkish peers from stigmatization, one main reason being the high rate of marginalization among them" (Pels and De Haan 2003, p. 72).However, students of Moroccan descent are doing better than their peers of Turkish descent or Antillean students in terms of their level of education and learning achievements (Centraal Bureau voor Statistiek 2018).Underlying socialisation patterns might explain these differences (Van Bergen et al. 2017).The extensive review study by Pels and De Haan (2003) shows that Moroccan parents in the Netherlands on average are more authoritarian than Dutch parents.Moreover, adolescents of Moroccan descent rely more heavily on parental advice than their Dutch peers.They also rely more on their peers for support, which leads to behaviour with which they distance themselves from Dutch society.The relation between socialisation, participation and citizenship knowledge deserves more attention, both quantitative and qualitative.
Where school type differences between students are concerned, we see that while there are differences in favour of students in higher school types, these are small; smaller than in the ICCS studies.This effect was the greatest in the domain of acting democratically.This finding seems to be related to the high external differentiation of the Dutch education system.Small-scale, qualitative research shows that, in the higher tracks, teachers discuss politics and the democratic constitutional state more often with their students than in the lower tracks (Nieuwelink et al. 2019).In order to identify causal relationships, however, a longitudinal research design is needed.
The differences in knowledge level that emerge from our study concern the effects found after controlling for school characteristics.Although we see clear school differences in student citizenship knowledge (especially concerning acting democratically), these differences cannot be explained by school size, the public or private character of the school, the degree of urbanisation or a more homogeneous or heterogeneous student population in terms of parental level of education.Parental education is strongly correlated with school type in the Dutch education system, however.Students with low-educated parents are overrepresented in prevocational education, while children of highly educated parents are found predominately in pre-university education.This is reflected in unequal student outcomes, not only with regard to academic achievement, but also in the field of citizenship (Van de Werfhorst 2014).Concerning the ethnic origin of students, even after controlling for, inter alia, school type, we see a relationship with student knowledge level in the domain of acting democratically.However, in order to be able to adequately estimate the (net) school effect, a more detailed check of the composition of the student population would be desirable.In our study, this has been achieved only to a certain extent by including variables that emerge from the literature as important factors at the student level (gender, ethnicity and SES).Since there was no exhaustive analysis-an adequate population estimation was not the aim of our analyses-some caution in the interpretation of the school effect is appropriate.
It is often argued that schools can compensate for inequalities in citizenship knowledge.In particular, it is alleged that schools can broaden horizons for those who have less positive experiences of politics and democracy.The literature shows that schools can indeed affect adolescents' knowledge of democracy, particularly by fostering an open classroom climate (Geboers et al. 2013) and by incorporating specific citizenship courses into the formal curriculum (Hooghe and Dassonneville 2011;Janmaat et al. 2014).Moreover, in schools where citizenship is regularly taught, the differences in citizenship knowledge between students of various social backgrounds are smaller than in schools where citizenship education is more or less neglected (e.g., Gainous and Martens 2012;Neundorf et al. 2016).However, studies also report that schools are often reproducing or strengthening existing differences rather than lowering them (Nieuwelink et al. 2019).The studies concerned do not, however, situate citizenship knowledge in the national societal and political context of students, nor do they differentiate between the political and social aspects of citizenship knowledge.An earlier study indicated that the pathways of citizenship competences of different groups of adolescents are related to the different aspects of citizenship (Geboers et al. 2015).Our study offers opportunities for investigating this in more depth for citizenship knowledge.This can help schools to meet their statutory obligation to promote active citizenship and to properly mould the compensatory effect that schools can have.However, to mitigate inequalities in citizenship knowledge, further research is necessary to investigate micro-level mechanisms within schools, such as level of teacher professionalism and teaching practices, that influence the various facets of student citizenship knowledge.Finally, the relationship with citizenship skills and attitudes also needs more attention.One of the crucial questions that have remained unanswered is to what extent supporting democratic values is a prerequisite to or the effect of citizenship knowledge.

Appendix A
We found empirical evidence for the convergence validity of the citizenship knowledge instrument.As presented in Table 1 there was a weak to average positive correlation between, on one hand, the average skill scores for citizenship knowledge (for citizenship knowledge in total and per social task) and, on the other hand, the average scores for citizenship attitude (range: r = 0.09 **-r = 0.31 **) and citizenship skill (range: r = 0.12 **-r = 0.24 **).Citizenship knowledge, however, showed no correlation or a weak negative correlation with the citizenship reflection scales (range: r = −0.14**-r = −0.01),apart from a weak positive correlation between citizenship knowledge and reflection on dealing with conflicts (range: r = 0.15 **-r = 0.19 **).
The results show in brief that the instrument meets the common requirements for validity and reliability, with correlation in the expected directions.This means we have an instrument that lends itself well for measuring knowledge of both the political and the social aspects of citizenship, based on the social "tasks" of acting democratically, acting in a socially responsible manner, dealing with conflicts and dealing with differences, which are a familiar part of the daily lives of young people.

Table 1 .
Conceptual framework of the citizenship knowledge instrument divided into the four social tasks and subtasks.

Table 2 .
Distribution of students from the representative sample of Dutch students in the third year of secondary school (n = 5163; 80 schools).
Notes: 1 The Dutch educational system consists of privately and publicly governed schools.Both types of schools are publicly funded.Private schools offer "denominational education", i.e., based on a specific religion or a particular educational philosophy. 2 Vmbo-basis: basic prevocational education; vmbo-kader: medium-level prevocational education; vmbo gemengd-theoretisch: lower general education; havo: higher general secondary education; vwo: pre-university education; mbo: intermediate vocational education.

Table 4 .
(A-E) Differences in citizenship knowledge according to student and school characteristics

Table A1 .
Correlations (r)between citizenship knowledge and other citizenship components: citizenship attitudes, citizenship skills and citizenship reflection.: 1 AD: Acting Democratically; ASR: Acting in a Socially Responsible manner; DC: Dealing with Conflicts; DD: Dealing with Differences; Attitude AD1: Acting Democratically Attitude: Willing to hear every voice; Attitude AD2: Willing to make critical contribution; Skill AD1: Able to state personal viewpoint; Skill AD2: Able to listen to other viewpoint; Skill ASR/DC: Skill Acting in a Socially Responsible manner and Dealing with Conflicts. 2** = p < 0.01, * = p < 0.05. Notes