Students’ citizenship competencies: The role of ethnic school composition and perceived teacher support

ABSTRACT

The current study examined the effects of school ethnic composition and teacher-student relationships (teacher support) on students’ citizenship competencies. Additionally, this study investigated the moderating effect of teacher support on the relationship between school ethnic composition and citizenship competencies. Citizenship was operationalized as competencies and knowledge concerning acting democratically, acting in a socially responsible manner, dealing with conflicts, and dealing with differences. Multilevel analyses among a sample of 4,902 students from 75 Dutch secondary schools showed that the degree of school ethnic diversity is positively related to competencies in acting democratically, acting in a socially responsible manner, and dealing with differences, but it is negatively related to knowledge regarding acting in socially responsible manner and dealing with conflicts. Teacher support was positively associated with both competencies and knowledge in all social tasks. Except for one outcome, no moderation effect of teacher support was found.

KEYWORDS:

• Citizenship competencies
• school diversity
• school ethnic composition
• secondary education
• teacher support

In the past decades, European countries have been facing challenges concerning increasing individualization and the erosion of civic participation and mutual involvement between citizens. As a result, legislation has been put in place in many countries to combat such adverse developments. Within this approach, a major role has been attributed to schools by the introduction of citizenship education as a legal task in many European countries (Eurydice, 2017; Hahn, 2020). In the Netherlands, this emphasis resulted in a legal task for schools aimed at the promotion of “active citizenship and social integration” (Dutch Ministry of Education, Culture and Science, 2005). The Dutch Education Council proposes that schools should endeavor to cultivate citizens who are able and willing to actively contribute to their communities and society at large (Dutch Education Council, 2003; Ten Dam & Volman, 2007). Accordingly, in this article we define citizenship not as a legal status per se, but as a general concept indicating people’s participation in and contribution to societal life. The capabilities that one needs to enact citizenship are referred to as citizenship competencies. In the Netherlands, where society is characterized by heterogeneity, it is additionally suggested that openness to diversity should be expected of citizens (Berlet et al., 2008). Therefore, emphasis is also placed on the importance of students developing the competencies to navigate diverse contexts (Dutch Ministry of Education, Culture and Science, 2005).

Although the effectiveness of schools as venues where citizenship is promoted has received substantial attention in the literature (for a review, see Geboers et al., 2013), previous research has largely neglected the school context within which students’ citizenship competencies are developed. In the Netherlands, the increasing diversity in society has resulted in more heterogeneous student populations, especially in urban areas. In general, it can be stated that in school—despite segregation—students are amid a more diverse population in comparison to their private networks (Eidhof, 2019; Parker, 2005, 2006; Vermeij et al., 2009).

For the purpose of this article, the make-up of the student population in terms of ethnic background is referred to as “school composition.” More specifically, the focus is on the degree of “school ethnic diversity,” which indicates the number of different ethnic groups (including native Dutch) in school and their relative size (Putnam, 2007). The degree of school ethnic diversity may influence the opportunities that students have to practice and learn about citizenship, as the student composition determines the chances of meeting others with different customs, cultures, and viewpoints (Parker, 2005). The first objective of this study is to add to the knowledge base on school composition and citizenship by investigating whether the make-up of the student body is related to students’ citizenship competencies.

As a second objective of the study, we intend to examine one characteristic of the multidimensional social school and classroom climate that may affect the relationship between school ethnic composition and students’ citizenship outcomes: the quality of teacher-student relationships (Berkowitz et al., 2017; Wang & Degol, 2016). In this study, the quality of the teacher–student relationship is conceptualized as the amount of perceived teacher support, referring to the degree to which students perceive that their teachers appreciate them and form personal relationships with them (Ryan & Patrick, 2001). Earlier research has shown a positive relationship between teacher-student relationships and student academic outcomes (Košir & Tement, 2014; Wang & Degol, 2016). It is important to investigate whether this positive association also holds true for students’ citizenship competencies given the concerns relating to social cohesion in society and schools’ tasks regarding citizenship.

Moreover, positive teacher-student relationships may moderate the effect¹ of school ethnic diversity on citizenship competencies. This reasoning is based on the finding that certain teaching characteristics may lead to different learning outcomes depending on students’ socio-cultural background. For instance, it has been observed that teacher quality seems to have a larger impact on learning outcomes in schools with high proportions of minority students and students from low socioeconomic status (SES) backgrounds (Payne, 2008; Presley et al., 2005). It could be conversely reasoned that the relationship between school ethnic composition and citizenship competencies differs according to the level of perceived teacher support. One could hypothesize that higher levels of perceived teacher support may buffer a possible negative effect or may intensify a possible positive effect of school composition on citizenship competencies.

Theoretical background

Young people’s citizenship and the school

Viewed conceptually, citizenship refers to the way in which members of a particular community or society organize living together and the way those members take part in societal life. In the Dutch context, like in many countries worldwide, the concept is guided by the principle of an “active” citizenry. It is expected that (young) people are willing and able to be part of a community and take the responsibility to actively contribute to that community (Dutch Education Council, 2012). Schools have an important role in nurturing such active citizens. In this article, we rely on the work by Ten Dam et al. (2011) to conceptualize citizenship competencies of young people. We contend that in preparing young people for societal life, the focus should be on young people’s lived experiences in their daily lives (cf. Biesta, 2007). Young people are learning and participating in society through daily social practices in which they interact with others within various contexts, including family and school settings, through (social) media use, and during leisure time (Lawy & Biesta, 2006). Accordingly, we follow Ten Dam et al.’s (2011)proposition that citizenship of young people manifests itself and is shaped in daily “social tasks” that they encounter. They are, for instance, faced with a social task when controversial issues are discussed in the classroom, when they see a person in need of help, or when they need to settle a conflict with a stranger. In short, young people’s experiences and their outlook on society take place and are shaped in their physically immediate and proximate surroundings, but this does not mean that the learning or teaching of citizenship is confined to the local or national context (Osler & Starkey, 2018).

For a long time, citizenship has been approached from a national perspective in citizenship curricula. However, in more recent years, educational literature has shifted its attention toward an approach of citizenship that has a more global orientation at its center (Byker, 2016; Estellés & Fischman, 2021; Goren & Yemini, 2017; Osler & Starkey, 2018). Comparable to the more traditional interpretations of citizenship, the exact meaning and contents of global citizenship are contested and present conflicting viewpoints (Byker & Putman, 2019; Estellés & Fischman, 2021; Goren & Yemini, 2017). However, a common viewpoint is that being a member of a community has moved beyond national borders as people are more interconnected across the globe (Byker, 2016; Osler & Starkey, 2018). This interconnectedness underscores the importance of shared humanity and being aware of and interested in global issues (Byker & Marquardt, 2016; Lilley et al., 2015). It is argued that, especially in Europe, a global perspective on citizenship (education) is urgent due to the context of challenging political and societal climates that stem from increasing immigration and intolerance (Osler & Starkey, 2018). These developments have ignited the need to prepare young people for their role in a complex global society and equip them with the knowledge and skills to be critical global citizens (Byker, 2016; Byker & Putman, 2019).

Byker (2013) has introduced Critical Cosmopolitan Theory as a framework for understanding critical consciousness of citizens within the context of a globalized world (Byker, 2016; Byker & Marquardt, 2016). This theory includes four central competencies: investigating the world, recognizing multiple perspectives, having the ability to communicate ideas, and taking action in societal issues (Byker, 2013; Byker & Marquardt, 2016). These competencies go hand-in-hand with the cultivation of critical consciousness as citizens start evolving from “reading” to “rewriting” the world (Byker, 2016; Byker & Marquardt, 2016). The former refers to citizens that are fully aware of diversity, pluralism, and social injustice, and the latter denotes responding to injustice by taking action (Byker, 2016; Byker & Marquardt, 2016; Byker & Putman, 2019).

Precisely in societies that are characterized by diversity and inequality, such as the Netherlands, great opportunities arise to connect the local to the global (cf. Byker & Marquardt, 2016). In fact, issues on a global scale have entered Dutch classrooms and triggered controversy regarding topics as diversity, politics, and social justice. In recent years, the influx of refugees from Syria, the attempted coup in Turkey, the killing of George Floyd in the United States, and the murder of teacher Samuel Paty in France have found traction in Dutch schools. In these schools—especially in the ones that are more diverse—students from various backgrounds meet each other. In such a context, international developments can spark controversy due to students having different allegiances and perspectives. However, it seems Dutch schools do not fully utilize this diversity yet, as an international comparative study shows that secondary education students in the Netherlands score lower on citizenship knowledge compared to countries that resemble the Netherlands (Munniksma et al., 2017). Moreover, students in the Netherlands, more so than students in comparable countries, attach great value to respecting the right to one’s own opinion. Lastly, in the Netherlands, students are less supportive of equal rights for men and women, ethnic minorities, and migrants compared to countries resembling the Netherlands.

In a societal climate that is characterized by diversity and its challenges, young people need to be equipped with competencies to navigate society actively and critically. Accordingly, citizenship of young people can be observed in “social tasks” that they execute in their daily lives (Ten Dam et al., 2011; Ten Dam & Volman, 2007). Ten Dam et al. (2011) distinguished four types of such social tasks: acting democratically, acting in a socially responsible manner, dealing with conflicts, and dealing with differences. A young person that acts democratically is, for instance, able to consider different viewpoints and express a well-informed opinion. Someone who acts in a socially responsible manner has the will to stand up for others and can adopt a socially just position. A young citizen who is able to deal with differences has knowledge of and is interested in (cultural) differences. Finally, being able to handle conflicts includes searching for a solution to conflicts and thinking about why a conflict arose. For a more detailed description of the social tasks, see the conceptual overview in Table 1. In short, young people’s citizenship is conceptualized broadly and comprehensively, reflecting a broad array of social situations with which they are confronted. Moreover, a parallel can be drawn between the social tasks presented by Ten Dam et al. (2011) and the four global competencies that belong to Critical Cosmopolitan Theory (Byker, 2013) in the sense that they both focus on preparing people to be critical citizens within diverse contexts.

Table 1. Conceptualization of citizenship competencies in terms of components and social tasks (table taken from Ten Dam et al., 2011)

COMPONENTS	Knowledge knowing, under- standing, insight	Attitudes thoughts, desires, willingness	Skills estimate of what one can do	Reflection contemplation of topics
SOCIAL TASK	A young person with such knowledge …	A young person with such attitudes …	A young person with such skills …	A young person with such reflection …
Acting democratically Acceptance of and contribution to a democratic society	… knows what democratic principles are and what acting in accordance with them involves.	… wants to hear everyone’s voice, enter into a dialogue and make an active, critical contribution.	… is able to assert own opinion and listen to the opinions of others.	… thinks about issues of democracy, power/ powerlessness, equal/unequal rights.
Acting in a socially responsible manner Taking shared responsibility for the communities to which one belongs.	… knows social rules (i.e., legal or unspoken rules for social interaction).	… wants to uphold social justice, is prepared to provide care and assistance, does not want to harm another or the environment as a result of his or her behavior.	… can adopt a socially just position.	… thinks about conflicts of interest, social cohesion, social processes, group processes (e.g., inclusion, exclusion), and own contribution to social justice.
Dealing with conflicts Handling of minor situations of conflict or conflicts of interest to which the child him/ herself is a party.	… knows methods to solve conflicts such as searching for win-win solutions, calling in help from others, admission of mistakes, prevention of escalation.	… is willing to explore conflicts, prepared to consider the standpoint of another, jointly searches for an acceptable solution.	… can listen to others, put oneself in someone else’s position, seek win-win solutions.	… thinks about how a conflict can arise, the role of others and oneself, and the possibilities to prevent or solve conflicts.
Dealing with differences Handling of social, cultural, religious, and outward differences.	… is familiar with cultural differences, has knowledge of rules of behavior in different social situations, knows when one can speak of prejudice or discrimination.	… has a desire to learn other people’s opinions and lifestyles, has a positive attitude toward differences.	… can adequately function in unfamiliar social situations, adjust to the desires or habits of others.	… thinks about the nature and consequences of the differences between people and cultural backgrounds for behavior and processes of inclusion and exclusion.

School composition and citizenship competencies

Being part of a school community is a double-layered phenomenon; first, students are practicing for citizenship, in the future or for communities outside school, and second, students are enacting citizenship, at present and in school. In school, students may encounter different lifestyles and perspectives (Parker, 2005); thus, the composition of the student population may play a role by providing opportunities and barriers to simultaneously enact and practice for citizenship.

If one conceptualizes students’ citizenship competencies based on the aforementioned daily social tasks, contradictory outcomes may be expected with regard to the influence of the school composition. This influence may depend on the mechanism that is assumed to be at play for a particular social task (Dijkstra et al., 2015). Following Dijkstra et al. (2015), we draw on functional community theory (Coleman & Hoffer, 1987) and the ecological development model (Bronfenbrenner, 1979). These frameworks imply that consistency and congruence between different social settings in which the person moves, such as home, the neighborhood and school, may lead to a more fruitful context for citizenship learning and development. Uniform contexts usually show higher degrees of social cohesion (Dijkstra et al., 2015). Consequently, a homogenous setting may be more conducive to the learning of certain aspects of citizenship, such as shared norms and conduct rules. On the other hand, heterogeneous environments may be more advantageous in other regards. Such environments could bolster out-group trust (Dijkstra et al., 2015). Moreover, they may offer more opportunities to gain access to new information, different attitudes, and different behavior options (Cochran, 1990). These outcomes result from the fact that in (non-familial) networks members come from a broad spectrum of different contexts (Cochran, 1990). In line with this observation, diversity within the school context may be beneficial for the opportunity to learn in cases of dissimilarities, which may apply to learning to handle differences and conflicts (Dijkstra et al., 2015), both of which are important aspects of citizenship competencies (see Ten Dam et al., 2011).

Previous research on school ethnic composition and students’ citizenship outcomes is often narrow in terms of its conceptualization of citizenship, or the measured concept is not defined explicitly as citizenship. Overall, prior studies are non-conclusive. Both positive (Janmaat, 2012) and negative (Campbell, 2007) associations have been reported, and a great deal of the findings also indicate non-relationships (e.g., Keating & Benton, 2013). The only study that we know of with a more comprehensive scope is a study by Dijkstra et al. (2015) which examined the association between school composition and pupils’ citizenship competencies in the final year of primary education. In their study, citizenship competencies are conceptualized as the knowledge, skills, attitudes, and reflection that young people need in order to function adequately in light of the social tasks that we discussed above (see Ten Dam et al., 2011). Dijkstra et al.’s (2015)findings revealed that school ethnic diversity was positively related to citizenship knowledge and reflection across the social tasks. In the current study, our objective is to add to the literature by using a similarly comprehensive conceptualization of citizenship competencies but with a focus on secondary education. Moreover, we differentiate between the social tasks rather than between knowledge, attitude, skills, and reflection across the social tasks. As the degree of heterogeneity of the school context may result in contrasting effects on citizenship competencies depending on the specific task that is measured, it seems more relevant to consider the contents of the social tasks instead of their components.

The quality of teacher-student relationships

A school community does not merely consist of dynamics within the student population but is also dependent on other features of the school social climate, such as teacher-student relationships (Wang & Degol, 2016). Teachers’ behavior toward students plays a critical role in the classroom learning environment (Baker, 1999). It has been found that warm and caring relationships between students and teachers contribute to an environment that promotes learning (Wang & Degol, 2016). Past studies have shown that favorable teacher-student relationship indicators—for instance, teacher support and trust—are positively associated with student achievement and motivation (e.g., Berkowitz et al., 2017; Cornelius-White, 2007; Goddard et al., 2001; Wubbels & Brekelmans, 2005) and negatively related with psychological and behavioral outcomes such as depression and misconduct (Wang et al., 2013). In short, a growing body of research has demonstrated that students benefit from teachers with whom they have positive relationships.

Similarly, we expect that high-quality teacher-student relationships are positively related to student outcomes in the area of citizenship. Flanagan (2013) argued that young people’s perception of society is shaped in indirect ways by their interactions in so called “mediating institutions” such as schools (Flanagan, 2013; Flanagan et al., 2007). Through their experiences in these small-scaled communities that are close to them, students create images of what it means to be a member of the more distant broader society. In such a context, teachers may serve as an example in their role as “proximate authority figures” (Flanagan, 2013; Flanagan et al., 2007) or “civic role models” (Campbell, 2008) and may be decisive in students’ development of citizenship skills. After all, teachers inevitably and continuously transfer values to students (Sandström Kjellin et al., 2010). Taken together, students’ ideas, knowledge, and attitudes about citizenship-related phenomena such as relations to authority, power structures, and attitudes toward society may be based, in part, on students’ interactions with closer authorities such as teachers (Cook, 1985; Flanagan, 2013; Flanagan et al., 2007). This finding coincides with the finding that receiving respect from and having a good relationship with teachers, promotes students’ view of schools as sites where opinions can be freely expressed and debated with those with differing opinions (Maurissen et al., 2018). Moreover, having a sense of being cared about by teachers stimulates students to engage in classroom activities (Wentzel, 1997).

The little research available in the domain of citizenship demonstrates that teacher-student relationships are—albeit weakly- positively related to attitudes toward conventional citizenship (e.g., voting) and social-movement related citizenship and, remarkably, negatively related to civic knowledge (Isac et al., 2014). In Isac et al.’s (2014)study, no association was found between teacher-student relationships and students’ future intention to participate in society. Thus, although two of the four outcomes in this study are positively associated with student-teacher relationships, the findings are still rather indistinct and puzzling. Two other studies have also found a positive link between teacher-student relationship and societal involvement (Wanders et al., 2020a, 2020b). Moreover, in a sample of immigrant students, Rutkowski et al. (2014) found that good student-teacher relationships are positively associated with involvement in the school and community, as well as trust in civic institutions. In our study, the objective is to investigate the association between teacher-student relationships and students’ citizenship competencies in the social tasks that young people encounter in their daily lives (see Ten Dam et al., 2011). As the literature points to a positive direction, a similar positive association may be expected between teacher-student relationships and our conceptualization of citizenship competencies.

Finally, we intend to investigate possible interactions between the school composition and teacher-student relationships. It has been previously stated that teacher quality may have more impact in schools where minority and low-SES students are overrepresented (Payne, 2008; Presley et al., 2005). Similarly, positive teacher-student relationships may play a more crucial role in heterogeneous schools and classes compared to less diverse ones. Conversely reasoned, the effects of school composition on citizenship competencies may depend on the levels of teacher support.

In this study, we explore whether a positive link between school ethnic diversity and citizenship competencies becomes stronger when teacher-student relationships are positive. Contrastingly, in the opposite direction, in the case of a negative association, good teacher-student relationships could provide mitigation or protection, reducing negative relationships between diversity and citizenship outcomes. Managing a classroom is a complex and challenging teaching activity, sometimes even more so in multicultural settings (Milner & Tenore, 2010; Van Tartwijk et al., 2009), as teachers have to navigate a variety of backgrounds, experiences, and perspectives. In such a multi-faceted context, establishing positive teacher-student relationships may have extra impact, either by further strengthening a positive relationship or by mitigating the possibly negative influence of high diversity. In other words, we will investigate whether the effects of a more diverse school composition could be respectively offset or enhanced by positive teacher-student relationships. In short, in this article, we address the question of whether school ethnic composition and teacher-student relationships are related to students’ citizenship competencies and, if so, whether there is an interaction between school composition and perceived teacher support in their relationship with students’ citizenship competencies.

Method

Research design

We used data collected in 2016 from a sample of Dutch secondary education schools that took part in the large-scale ESC research project (Understanding the Effects of Schools on Students’ Citizenship). This project investigates citizenship education in secondary schools and 9^th grade students’ competencies in the citizenship domain (also, see Sincer et al., 2021). In total, 82 schools confirmed their participation in the study. These schools were recruited in two ways: by employing a random sampling procedure (n = 52)² and through research team members’ social networks (to increase statistical power, n = 30). In the sample, secondary education schools across the Netherlands were largely represented, with public schools and schools in the North-Holland province slightly overrepresented. Schools providing only lower-level pre-vocational education were slightly underrepresented. One of the schools ceased participation at the initial stage of the study for unknown reasons. Therefore, ultimately, the sample consisted of 81 secondary schools, 240 classes, and 5,297 students.

Data for the overall ESC project were collected through anonymous online questionnaires, filled out by 9^th grade students, teachers, team leaders, and school leaders. For the present study, we used student data. Schools received written instructions on how to randomly select three 9^th grade classes. Prior to administering the questionnaires, students’ parents received a letter with an explanation of the study aims and procedures, and they could reject their child’s participation. Students filled out two questionnaires in two regular classes, during which a trained test leader was present.

The first questionnaire contained questions about student background characteristics and school characteristics referring to school climate and/or citizenship education. Additionally, for practical reasons, the first questionnaire contained a few items from the Citizenship Competencies Questionnaire (CCQ, for an extensive description, see Ten Dam et al., 2011). The second questionnaire consisted of the vast majority of the CCQ, next to newly constructed knowledge items (see below for a description).

Several criteria were set up for the inclusion of students, classes, and schools in the analysis: a) students had to have completed most of the relevant items on student background variables, as well as all items on teacher support and the CCQ; b) a minimum of 10 students per class had to have filled out the questionnaires; c) a minimum participation rate of 60% per class was required, and d) based on criteria b and c, only schools with a maximum of one eliminated class were included in the analysis. Based on these criteria, three schools were excluded from the analysis. Additionally, two schools that shared the same school building were also excluded from the analysis, as there was no clarity on their degree of interrelatedness. A third school did not match the characteristics of the overall sample due to having a high share of special needs students and was, therefore, omitted from the analysis. Altogether, data were available from 4,942 students from 75 schools. Additionally, potential multivariate outliers were investigated. In this article we report the findings excluding the cases containing multivariate outliers (n = 40), resulting in a total sample of 4,902 students.

Variables

Outcome variables: Students’ citizenship competencies

We used the CCQ to measure students’ citizenship competencies. The CCQ has been developed for young people between the age of 11 and 16, comprising items that correspond with young people’s daily citizenship practices. These practices are operationalized as the following four social tasks: acting democratically, acting in a socially responsible manner, dealing with conflicts, and dealing with differences. The competencies needed for each of these social tasks consist of four components: knowledge, attitude, skills, and reflection. Rather than being a task-based assessment, the CCQ measures students’ self-assessment concerning social tasks that they encounter daily (except for knowledge, which is measured through a test). Table 1 provides an overview of the conceptualizations of the social tasks and the components of the citizenship competencies they require. For the ESC study, we constructed new citizenship knowledge items to ensure a more balanced distribution of item difficulty levels and to add the possibility of differentiating between the separate social tasks. For each social task we incorporated attitude, skills, and reflection into a single construct of citizenship competencies. Knowledge is a different kind of component and is, therefore, operationalized separately (see Ten Dam et al., 2011 for a more detailed explanation).³ The method of combining the three components was chosen for three reasons. First, bringing together the three components was preferred for the purpose of parsimony; otherwise, 16 outcome variables would have been needed to be included and reported. Second, as stated previously, we reasoned that school composition may have differential effects depending on the particular social task that is being measured. Consequently, we were interested in whether school composition would show any effect on the social task scores, rather than on the distinct competencies. Third, it can be argued that the three components together reflect a comprehensive array of student dispositions in the affective-behavioral domain of citizenship (Schulz et al., 2016).

Although it may have been possible to select a single measure instead of combining all three components, we dismissed the first option as it would have been too narrow a measure of students’ overall competencies. Following Ten Dam et al. (2011), we fitted a second-order factor model to examine whether combining attitude, skills, and reflection (henceforth Competence) for each social task would be appropriate. The model fits were acceptable as for all social tasks it was found that TLI ≥ .90, CFI ≥ .91 and RMSEA < .08 (Van de Schoot et al., 2012). The second-order factor model justifies the assumption that there is indeed an underlying construct for the three components together. Therefore, for each social task, we created one score for Knowledge and one for Competence. The calculation of each score was based on multiple items (see Appendix A for the number of items). The Knowledge scores were calculated based on item response theory analysis (IRT), and the Competence scores were determined by averaging the scores on the corresponding items. To measure Knowledge, a multiple-choice test containing three response options was used. For the measurement of Competence, survey items were used that were scored with 4-point Likert scales. See Appendix A for Cronbach’s alphas, mean scores, and standard deviations of the components of the social tasks, and see Appendix B for item examples.

Furthermore, we executed multigroup factor analyses to determine whether valid group comparisons could be made (Van de Schoot et al., 2012). Ethnic background (migrant and native background) and educational track (pre-vocational and other track) were included for testing measurement invariance of attitude, skills, and reflection for all social tasks. Configural, metric, and scalar invariance were sequentially tested in MPlus. In all models ΔCFI (configural compared to metric and metric compared to scalar) stayed within the 0.01 threshold (Cheung & Rensvold, 2002). Thus, it was concluded that group comparisons using this model are valid comparisons. The same holds for the knowledge component of the social tasks, based on a measurement invariance test conducted in jMetrik 4.0.5 (Meyer, 2014).

Individual-level variables (control variables)

Control variables were added to the analyses, both on the individual level and the school level. Some of these variables have been previously demonstrated to be related to certain citizenship outcomes (e.g., Munniksma et al., 2017; Schulz et al., 2010; Wanders et al., 2020a, 2020b). On the student-level, the control variables were as follows. Migrant background was defined as a dichotomous variable (1 = migrant background, 0 = native) based on information from the student questionnaire. Following Statistics Netherlands (2016)definition, if either of the parents was born outside the Netherlands, the student was classed as having a migrant background. In case of missing data on the parents, the student’s country of birth was used for the classification. Approximately a quarter of the students in our sample had a migrant background (23.9%).

Parents’ educational level served as an indicator for students’ socioeconomic status (SES). Based on information provided by the student, SES was determined according to the parent with the highest educational level. It has been previously found that there is a positive association between the educational level of parents and citizenship outcomes (Schulz et al., 2010; Wanders et al., 2020a). This result may stem from a richer learning environment provided by high-educated parents (Wanders et al., 2020a). Accordingly, we conversely expected students from low-educated families to be more disadvantaged in terms of citizenship. Consequently, we created a dummy variable for SES (1 = low SES, 0 = not-low SES). Educational levels up to secondary vocational education were assigned to the low-SES group. Low-SES students constituted 11.7% of our sample. Furthermore, a variable for missingness on SES was created (1 = missing, 0 = not missing) in order to include students in the analysis for whom SES was not known (18%).

Finally, gender (male = 1, female = 0) and educational track (1 = pre-vocational track, 0 = other track) were added as control variables on the individual level. In our sample, 51.7% of the respondents were female. The pre-vocational track students accounted for 42% of the sample.

The teacher-student relationship

In this study, the quality of the teacher–student relationship is measured as the perceived amount of teacher support. As stated earlier, we conceptualized teacher support as the degree to which students feel they are cared about by, and form connections, with their teachers (Ryan & Patrick, 2001). To measure students’ perception of teacher support, we used a scale adapted from Malecki and Elliott (1999). Their original Student Social Support Scale (SSSS) consists of four subscales that tap support from different sources, one of which are teachers. The scale constitutes different types of support, including emotional support. We used an adapted version of this subscale as it is in harmony with our conceptualization of teacher support, and the SSSS has been proven to be a valid and reliable measurement (Malecki & Elliott, 1999). The adapted scale consists of eight items with response options ranging on a 5-point Likert scale (from “strongly disagree” to “strongly agree”). The items are: “My teachers care about me,” “My teachers treat me in a good and fair way,” “My teachers try to answer my questions,” “My teachers help me,” “My teachers understand me,” “My teachers take the time to talk to me about what’s important to me,” “My teachers compliment me when I’ve done something right,” and “My teachers listen to me when I have a problem.” For the scale, construction responses were averaged (Cronbach’s alpha = .87). In our sample, the average score was 3.65 (SD = .57).

School-level variables

School ethnic composition was operationalized as the degree of school ethnic diversity. To this end, following previous studies, a Herfindahl Index was calculated (Dijkstra et al., 2015; Lancee & Dronkers, 2011; Putnam, 2007). This index indicates the total number and size of ethnic groups.⁴ To illustrate, a completely homogeneous environment has an index of 0, whereas an entirely diverse context results in an index of 1. To calculate the index, students were assigned to nine large ethnic groups depending on parents’ country of birth. In case of both parents being born abroad, mothers’ country of birth was used for the classification. Missing information on one parent led to using information of the other parent. If no information was available on either parent, we considered the student’s country of birth. We constructed the following nine groups: Native-Dutch (75.9%), Turkish (4.0%), Moroccan (5.1%), former Dutch colonies (Surinam, Netherlands Antilles, and Aruba, 2.9%), European (North, West, South, 3.3%), Mid- and Southeast European and other Western (2.0%), Middle Eastern (1.4%), Asian (2.7%), Sub-Saharan African and other Non-Western (2.8%). Creating more specific groups was not possible due to the small numbers of students within certain groups. On average, the diversity index in our sample was .34 (SD = .23).

We included school SES composition, school size, degree of urbanization, and school type as control variables. SES composition was constructed by aggregating individual SES to the school level by calculating the proportion of students with a low-SES background. The average proportion of students with low-SES in schools was .12 (SD = .08). We obtained information on the number of students per school location (school size) from a dataset of the Education Executive Agency (2015). The mean school location size was 872 (SD = 529).

The degree of urbanization of the school location was operationalized according to data from Statistics Netherlands (2012). Five categories were classified (1 = rural, 5 = highly urban) using the number of home addresses per postal code area. In more urban locations, people may be challenged more in terms of enacting and practicing for citizenship due to the complexities of the environment (e.g., Geijsel et al., 2012; Zwaans et al., 2008). In our sample, the mean degree of urbanization was 3.56 (SD = 1.30).

Finally, schools were classified based on their school type. Dutch schools are either “categorial,” meaning that the school provides education for a specific track (e.g., for the pre-vocational track), or “comprehensive,” which means that more than one track is available. It could be argued that students in the categorial pre-vocational schools are potentially more disadvantaged concerning learning opportunities for citizenship, as they are both in a lower educational track and are isolated from students from other tracks. Pre-vocational categorial schools accounted for 32% of the sample.

Analyses

To account for the hierarchical data structure (students in schools), multivariate multilevel analyses were carried out using Mplus 8.1 (Hox, 2002; Snijders & Bosker, 1999). When analyzing multiple dependent variables, as is the case in this study, it is most suitable to conduct multivariate multilevel analysis in which all outcome variables are included concurrently. This type of analysis increases statistical power and results in lower chances of a Type I error (Hox, 2002). Notwithstanding, in our model, the number of parameters outnumbered the number of schools, increasing the risk of a somewhat unreliable analysis. Consequently, we performed multivariate multilevel analyses per social task. Thus, per social task, we included Competence and Knowledge simultaneously in the analysis.

The following variables were added to the models at the school level: the predictor variable school ethnic diversity and the control variables school SES composition, school type, school size, and degree of urbanization. At the individual level, we included teacher support as a predictor variable, and the following variables were added as control variables: educational track, gender, low-SES, SES missingness, and ethnic background. Grand mean centering was applied for school ethnic diversity, teacher support, and their interaction term.

In the first step, no predictor variables were included in the model, also known as the null model. This step was taken to ensure the appropriateness of the multilevel analysis. In Model 1, the control variables from both the individual and school level were added. In Model 2, school ethnic diversity was included to investigate its association with the outcome variables. Next, Model 3 included teacher support to examine its main effect on the outcome variables. Lastly, in Model 4, the cross-level interaction between school ethnic composition and teacher support was added. The next section provides the results of the analyses.

Results

Tables 2 and 3 show the correlations between the variables on the individual and school level. As part of the multilevel analyses, we compared model fits of the successive models. The model fits significantly improved in each consecutive step for all models (also see Tables 3 to 6). Additionally, for all significant results, effect sizes were calculated (see Appendix 3). The effect sizes range from 0.10 to 0.47, indicating small to moderate effects.

Table 2. Correlations between individual level variables (n = 4,902)

Measures	1.	2.	3.	4.	5.	6.	7.	8.	9.	10.	11.	12.	13.	14.
(1) SES	1
(2) SES missing	−.17**	1
(3) Ethnic background	.01	.12**	1
(4) Educational track	.14**	.19**	.08**	1
(5) Gender	.001	.04**	−.02	.05**	1
(6) Teacher support	−.009	−.01	−10**	.02	−.06**	1
(7) Act. democratically-Competence	−.04*	−.06*	.14**	−.07**	−.05**	.16**	1
(8) Act. democratically-Knowledge	−.12**	−.14**	−.16**	−.42**	−.13**	.14**	.12**	1
(9) Act. socially respons.- Competence	−.02	−.06**	.09**	−.07**	−.23**	.19**	.69**	.09**	1
(10) Act. socially respons.- Knowledge	−.07**	−.11**	−.14**	−.30**	−.19**	.16**	.07	.55**	.07**	1
(11) Dealing with conflicts- Competence	−.04**	−.09**	.02	−.13**	−.20**	.24**	.60**	.20**	.75**	.20**	1
(12) Dealing with conflicts- Knowledge	−.08**	−.10**	−.10**	−.34**	−.20**	.14**	.07**	.54**	.09**	.51**	.21**	1
(13) Dealing with differences- Competence	−.06**	−.09**	.17**	−.09**	−.21**	.10**	.64**	.09**	.70**	.08**	.58**	.09**	1
(14) Dealing with differences- Knowledge	−.10**	−.14**	−.07**	−.37**	−.17**	.13**	.12**	.60**	.09**	.52**	.20**	.55**	.11**	1

Note. Pearson’s correlation coefficients are reported for the correlation between continuous and dichotomous variables and Phi coefficients are reported for the correlation between dichotomous variables.

*p ≤ .05. **p ≤ .01.

Table 3. Correlations between school level variables (n = 75)

Measures	1	2	3	4	5
(1) Ethnic school diversity	1
(2) School size	−.03	1
(3) SES composition	−.08	−.38**	1
(4) Degree of urbanization	.58**	.25*	−.31**	1
(5) School type	.07	−.58**	.40**	−.14	1

*p ≤ .05. **p ≤ .01.

Acting democratically: Competence and Knowledge

Table 4 shows the results of the stepwise multilevel analysis concerning acting democratically. School ethnic diversity demonstrated a significant positive relationship with Competence (β = .11). Thus, the more diverse the school, the more competent students feel in acting democratically. However, no significant relationship was found with Knowledge (β = −.28), which means that school ethnic diversity does not play a role in students’ knowledge in the area of acting democratically. Teacher support was positively associated with both Competence (β = .12) and Knowledge (β = .24). In other words, students who perceive to receive more teacher support report higher levels of Competence and Knowledge in acting democratically. Lastly, there was no significant cross-level interaction between school ethnic diversity and teacher support for either Competence or Knowledge. Stated differently, the degree of teacher support did not change the relationship between school ethnic diversity and citizenship outcomes. The analyses additionally showed that 6% of the variation in Competence scores and 27% of the variation in Knowledge scores can be attributed to differences between schools.

Table 4. Associations between school ethnic composition, teacher support and acting democratically (Competence & Knowledge)

		Competence				Knowledge
	Model 1	Model 2	Model 3	Model 4	Model 1	Model 2	Model 3	Model 4
Intercept	2.71 (.03)	2.72 (.04)	2.70 (.04)	2.70 (.04)	1.96 (.24)	1.82 (.23)	1.77 (.22)	1.77 (.22)
School level
Ethnic composition (Herfindahl index)		.05 (.04)	.11 (.04)**	.11 (.04)**		−.38 (.21)	−.26 (.21)	−.28 (.21)
School size	.00 (.00)**	.00 (.00)**	.00 (.00)**	.00 (.00)**	.00 (.00)	.00 (.00)	.00 (.00)	.00 (.00)
Degree of urbanization	.03 (.01)**	.03 (.01)**	.03 (.01)**	.03 (.01)**	−.04 (.03)	.01 (.04)	.00 (.04)	.00 (.04)
School type (pre-vocational)	−.01 (.03)	−.01 (.03)	−.01 (.03)	−.01 (.03)	−.10 (.12)	−.10 (.12)	−.10 (.12)	−.10 (.12)
SES composition (proportion low)	−.39 (.11)**	−.39 (.11)**	−.36 (.11)**	−.36 (.11)**	−1.80 (.74)*	−1.75 (.71)*	−1.69 (.68)*	−1.66 (.68)*
Cross-level
Ethnic school composition * Teacher support				n.a.				.10 (.17)
Individual level
Gender	−.04 (.01)**	−.04 (.01)**	−.03 (.01)	−.03 (.01)	−.26 (.04)**	−.26 (.04)**	−.24 (.04)**	−.24 (.04)**
Ethnicity (migrant)	.10 (.02)**	.09 (.02)**	.09 (.02)**	.09 (.02)**	−.24 (.04)**	−.23 (.04)**	−.22 (.04)**	−.22 (.04)**
Pre-vocational track	−.02 (.02)	−.02 (.02)	−.02 (.02)	−.02 (.02)	−.72 (.09)**	−.72 (.09)**	−.72 (.08)**	−.72 (.09)**
SES low	−.03 (.02)	−.03 (.02)	−.03 (.02)	−.03 (.02)	−.17 (.05)**	−.17 (.05)**	−.17 (.05)**	−.17 (.05)**
SES missing	−.07 (.01)**	−.07 (.01)**	−.06 (.01)**	−.06 (.01)**	−.10 (.04)*	−.10 (.05)*	−.10 (.05)*	−.10 (.05)*
Teacher support			.12 (.01)**	.12 (.01)**			.24 (.03)**	n.a.
Improvement fit (TRd)	484.82	6.71	182.08	12.01
Δdf	20	2	2	2

Notes. Standard errors in parentheses.

ICC’s of the null model for respectively Competence and Knowledge: .06 and .27.

*p ≤ .05. **p ≤ .01.

Acting in a socially responsible manner: Competence and Knowledge

The results for acting in a socially responsible manner are presented in Table 5. It shows that school ethnic diversity is significantly positively related to Competence (β = .11) and significantly negatively related to Knowledge (β = −.52). Thus, students in more diverse schools seem to be better at acting socially responsible, but at the same time, they score lower on knowledge in this area. In addition, there was a positive association between teacher support and both Competence (β = .14) and Knowledge (β = .29). Put differently, students who feel supported by their teachers are better able to act socially responsible and also show more knowledge in this domain. A negative cross-level interaction effect between the degree of school ethnic diversity and teacher support was found only for Competence (β = −.13). This result indicates that the more teacher support students perceive to receive, the less they benefit from being in a more diverse school. Thus, teacher support seems to compensate for a lack of diversity. The school level accounted for 4% and 17% of the variation, respectively, in Competence and Knowledge scores.

Table 5. Associations between school ethnic composition, teacher support and acting in a socially responsible manner (Competence & Knowledge)

		Competence				Knowledge
	Model 1	Model 2	Model 3	Model 4	Model 1	Model 2	Model 3	Model 4
Intercept	2.85 (.04)	2.87 (.04)	2.84 (.04)	2.84 (.04)	2.18 (.23)	1.95 (.21)	1.89 (.21)	1.90 (.21)
School level
Ethnic composition (Herfindahl index)		.06 (.04)	.12 (.04)**	.11 (.04)**		−.64 (.24)**	−.49 (.24)*	−.52 (.24)*
School size	.00 (.00)**	.00 (.00)**	.00 (.00)*	.00 (.00)*	.00 (.00)	.00 (.00)	.00 (.00)*	.00 (.00)
Degree of urbanization	.02 (.01)**	.02 (.01)*	.02 (.01)*	.02 (.01)	−.07 (.03)	.00 (.04)	.00 (.04)	.00 (.04)
School type (pre-vocational)	−.01 (.03)	−.01 (.03)	−.01 (.03)	−.01 (.03)	−.09 (.12)	−.07 (.12)	−.08 (.12)	−.07 (.12)
SES composition (proportion low)	−.28 (.11)*	−.28 (.11)*	−.24 (.11)*	−.24 (.12)*	−1.28 (.76)	−1.21 (.68)	−1.13 (.65)	−1.16 (.67)
Cross-level
Ethnic school composition * Teacher support				−.13 (.05)*				−.00 (.17)
Individual level
Gender	−.17 (.01)**	−.17 (.01)**	−.16 (.01)**	−.16 (.01)**	−.47 (.04)**	−.47 (.04)**	−.45 (.04)**	−.45 (.04)**
Ethnicity (migrant)	.05 (.01)**	.05 (.02)**	.05 (.02)**	.05 (.02)**	−.18 (.05)**	−.15 (.05)**	−.15 (.05)**	−.15 (.05)**
Pre-vocational track	−.02 (.02)	−.02 (.02)	−.02 (.02)	−.02 (.02)	−.56 (.10)**	−.56 (.09)**	−.55 (.09)**	−.56 (.09)**
SES low	−.02 (.02)	−.02 (.02)	−.02 (.02)	−.01 (.02)	−.09 (.05)	−.09 (.05)	−.09 (.05)	−.09 (.05)
SES missing	−.06 (.01)**	−.06 (.01)**	−.05 (.01)**	−.05 (.01)**	−.09 (.05)	−.09 (.05)	−.08 (.05)	−.08 (.05)
Teacher support			.14 (.01)**	n.a.			.29 (.04)**	n.a.
Improvement fit (TRd)	563.70	10.71	183.93	41.76
Δdf	20	2	2	4

Notes. Standard errors in parentheses.

ICC’s of the null model for respectively Competence and Knowledge: .04 and .17

*p ≤ .05. **p ≤ .01.

Dealing with conflicts: Competence and Knowledge

Table 6 presents the results for dealing with conflicts. School ethnic composition showed no significant association with Competence (β = .04). A significantly negative relationship was found between the degree of school ethnic diversity and Knowledge (β = −.43). Thus, students in more diverse schools have less knowledge on dealing with conflicts. Again, it was found that the more teacher support students perceived to receive, the higher their levels of Competence (β = .19) and Knowledge (β = .26). Finally, there was no cross-level interaction between school ethnic diversity and teacher support. Four percent of the variance in Competence and 17% of the variance in Knowledge could be attributed to school differences.

Table 6. Associations between school ethnic composition, teacher support and dealing with conflicts (Competence & Knowledge)

		Competence				Knowledge
	Model 1	Model 2	Model 3	Model 4	Model 1	Model 2	Model 3	Model 4
Intercept	3.05 (.04)	3.04 (.05)	3.00 (.04)	3.00 (.04)	2.07 (.21)	1.86 (.19)	1.81 (.19)	1.81 (.19)
School level
Ethnic composition (Herfindahl index)		−.04 (.04)	.05 (.04)	.04 (.04)		−.56 (.22)*	−.43 (.22)*	−.43 (.22)*
School size	.00 (.00)	.00 (.00)	.00 (.00)	.00 (.00)	.00 (.00)	.00 (.00)	.00 (.00)	.00 (.00)
Degree of urbanization	.01 (.01)	.02 (.01)	.01 (.01)	.01 (.01)	−.04 (.03)	.02 (.03)	.02 (.03)	.02 (.03)
School type (pre-vocational)	−.01 (.03)	.01 (.03)	.01 (.03)	.00 (.03)	−.15 (.11)	−.13 (.11)	−.14 (.11)	−.14 (.11)
SES composition (proportion low)	−.27 (.14)**	−.27 (.14)*	−.21 (.13)	−.22 (.13)	−1.39 (.67)*	−1.33 (.60)*	−1.25 (.58)*	−1.26 (.58)*
Cross-level
Ethnic school composition * Teacher support				−.06 (.06)				n.a.
Individual level
Gender	−.16 (.02)**	−.16 (.02)**	−.15 (.02)**	−.15 (.02)**	−.48 (.04)**	−.48 (.04)**	−.46 (.04)**	−.46 (.04)**
Ethnicity (migrant)	.02 (.02)	.03 (.02)	.03 (.02)	.03 (.02)	−.11 (.05)*	−.08 (.05)	−.07 (.05)	−.07 (.05)
Pre-vocational track	−.07 (.02)**	−.07 (.02)**	−.08 (.02)**	−.07 (.02)**	−.58 (.10)**	−.58 (.10)**	−.58 (.09)**	−.58 (.09)**
SES low	−.03 (.02)*	−.03 (.02)*	−.03 (.01)*	−.03 (.02)*	−.08 (.05)	−.08 (.05)	−.07 (.05)	−.07 (.05)
SES missing	−.07 (.01)**	−.07 (.01)**	−.07 (.01)**	−.07 (.01)**	−.03 (.05)	−.03 (.05)	−.03 (.05)	−.03 (.05)
Teacher support			.19 (.02)**	n.a.			.26 (.03)**	.26 (.03)**
Improvement fit (TRd)	595.04	10.74	303.91	16.90
Δdf	20	2	2	2

Notes. Standard errors in parentheses.

ICC’s of the null model for respectively Competence and Knowledge: .04 and .17

*p ≤ .05. **p ≤ .01.

Dealing with differences: Competence and Knowledge

Table 7 summarizes the results for dealing with differences. The results showed that school ethnic diversity is significantly positively related to Competence (β = .21); students in diverse schools assess themselves to be more competent in dealing with differences. No significant relationship was found between degree of school diversity and Knowledge (β = −.27). In accordance with the other social tasks, it was found that teacher support is positively associated with both Competence (β = .10) and Knowledge (β = .26). Again, no cross-level interaction effect between school ethnic diversity and teacher support was found. The school level accounted for 8% and 20% of the variation in respectively Competence and Knowledge scores.

Table 7. Associations between school ethnic composition, teacher support and dealing with differences (Competence & Knowledge)

	Competence				Knowledge
	Model 1	Model 2	Model 3	Model 4	Model 1	Model 2	Model 3	Model 4
Intercept	2.66 (.04)	2.72 (.04)	2.70 (.04)	2.70 (.04)	1.83 (.22)	1.69 (.20)	1.64 (.20)	1.63 (.20)
School level
Ethnic composition (Herfindahl index)		.17 (.04)**	.22 (.04)**	.21 (.05)**		−.38 (.23)	−.26 (.23)	−.27 (.23)
School size	.00 (.00)**	.00 (.00)*	.00 (.00)*	.00 (.00)*	.00 (.00)	.00 (.00)	.00 (.00)	.00 (.00)
Degree of urbanization	.04 (.01)**	.02 (.01)**	.02 (.01)**	.02 (.01)**	−.01 (.03)	.03 (.03)	.02 (.03)	.03 (.03)
School type (pre-vocational)	−.02 (.03)	−.02 (.03)	−.03 (.03)	−.03 (.03)	−.05 (.12)	−.04 (.12)	−.05 (.11)	−.05 (.11)
SES composition (proportion low)	−.33 (.14)*	−.35 (.14)*	−.32 (.14)*	−.31 (.15)*	−1.50 (.71)*	−1.46 (.66)*	−1.39 (.64)*	−1.33 (.63)*
Cross-level
Ethnic school composition * Teacher support				−.08 (.06)				.06 (.15)
Individual level
Gender	−.17 (.02)**	−.17 (.02)**	−.16 (.02)**	−.16 (.02)**	−.38 (.04)**	−.38 (.04)**	−.36 (.04)**	−.36 (.04)**
Ethnicity (migrant)	.13 (.02)**	.12 (.02)**	.12 (.02)**	.12 (.02)**	−.03 (.05)	−.02 (.05)	−.01 (.05)	−.00 (.05)
Pre-vocational track	−.03 (.02)	−.02 (.02)	−.03 (.02)	−.02 (.02)	−.71 (.10)**	−.71 (.10)**	−.71 (.10)**	−.72 (.09)**
SES low	−.08 (.02)**	−.08 (.02)**	−.07 (.02)**	−.07 (.02)**	−.18 (.05)**	−.18 (.05)**	−.18 (.05)**	−.18 (.05)**
SES missing	−.11 (.02)**	−.11 (.02)**	−.10 (.01)**	−.10 (.01)**	−.15 (.04)**	−.15 (.04)**	−.14 (.04)**	−.15 (.04)**
Teacher support			.10 (.02)**	n.a.			.26 (.04)**	n.a.
Improvement fit (TRd)	617.62	15.81	101.42	33.55
Δdf	20	2	2	4

Notes. Standard errors in parentheses.

ICC’s of the null model for respectively Competence and Knowledge: .08 and .20

*p ≤ .05. **p ≤ .01.

Lastly, for the sake of clarity, Table 8 presents a summary of the significant findings.

Table 8. Overview of significant findings

Relationship	Direction of relationship
Degree of school ethnic diversity &
Acting democratically: Competence Acting soc. responsible: Competence Dealing with differences: Competence	+ + +
Degree of school ethnic diversity & Acting soc. responsible: Knowledge Dealing with conflicts: Knowledge	- -
Teacher support & Acting democratically: Competence Acting soc. responsible: Competence Dealing with differences: Competence Dealing with conflicts: Competence	+ + + +
Teacher support & Acting democratically: Knowledge Acting soc. responsible: Knowledge Dealing with differences: Knowledge Dealing with conflicts: Knowledge	+ + + +
Interaction: Teacher support * school ethnic diversity -> acting socially responsible: Competence	-

Note. Positive relationships are denoted with a ‘+’ sign and negative relationships

are denoted with a ‘-‘ sign

Discussion and conclusion

In this study, our aim was to gain understanding of the association between the degree of school ethnic diversity, teacher support, and students’ citizenship competencies. To this end we, first investigated the extent to which school ethnic composition and perceived teacher support were related to students’ citizenship competencies, and then, we examined whether school ethnic composition and perceived teacher support interacted in their association with students’ citizenship competencies. Students’ citizenship outcomes were operationalized in terms of their scores on Competencies and Knowledge concerning four social tasks that represent young people’s daily life citizenship reality (cf. Ten Dam et al., 2011).

Concerning the direct relationship between school composition and citizenship competencies, our analyses showed that school ethnic composition is significantly positively related to competencies in acting democratically, acting in a socially responsible manner, and dealing with differences. Thus, the more ethnically diverse the school, the higher the levels of competencies in these three domains. The effect was strongest for dealing with differences. There was no relationship between school composition and competencies in dealing with conflicts.

The positive relationships confirm the assumption that, in comparison to less or non-diverse schools, students in diverse schools experience more encounters in which they can practice the social tasks (cf. Agirdag et al., 2016; Geijsel et al., 2012). This opportunity may either enhance students’ actual capacity to function adequately or it may lead to more positive self-assessments. It is also possible that both reasonings are valid. These reasonings also correspond with the largest effect size that we found for the relationship between school diversity and competencies in dealing with differences, as differences between students may be the most salient phenomenon that students encounter in diverse schools.

The explanation of the lack of relationship between school composition and competencies in dealing with conflicts may be sought in the nature of this particular social task. Estimating one’s own ability to deal with conflicts requires the imagination of a situation in which a negative condition has emerged. More importantly, in conflict situations, one’s personal interests are at stake or jeopardized more directly compared to situations relating to the other social tasks. Therefore, showing the competence to adequately deal with conflicts may be a challenge for all students, regardless of the school composition. Consequently, it is possible that schools in general, thus independent of the school composition, set learning to deal with conflicts as a priority.

When it comes to ethnic composition in relation to knowledge, no association was found with acting democratically and dealing with differences, while a negative relationship was demonstrated with acting in a socially responsible manner and dealing with conflicts. In other words, students enrolled in diverse schools have less knowledge on acting in a socially responsible manner and dealing with conflicts. Against the background of the diverse character of society at large, it may perhaps be the case that in all schools, irrespective of the student body, more focus is placed on the discussion and teaching of democratic issues and dealing with diversity. These topics may be more urgent in schools given the earlier mentioned tense societal climate in Dutch society. This interpretation may explain the non-effect of student composition on knowledge in the aforementioned areas, but it does not explain the negative relationship with knowledge of acting in a socially responsible manner and dealing with conflicts. Furthermore, it should be mentioned that the relationship between the degree of ethnic diversity and knowledge of acting democratically and dealing with differences also pointed in a negative direction. However, these relationships were not significant. Altogether, the findings may also suggest that ethnically diverse schools direct their attention more toward the development of competencies rather than to knowledge, resulting in lower scores on knowledge for students in more diverse schools.

In line with the above, what stands out in these findings are the differential effects on competencies and knowledge within and across the social tasks. Although it is difficult to interpret this observation content-wise per social task, it does coincide with the finding that the knowledge component of citizenship is of a different kind compared to its other components (Ten Dam et al., 2011). Our findings are, therefore, also another validation for the distinct analysis of Competencies and Knowledge. It is noteworthy that we additionally found a difference between competencies and knowledge in the degree to which the variation in scores could be attributed to school differences; the school-level accounted for a small amount of the variation (4–8%) in competence scores while this amount was higher for knowledge scores (17–27%). This result indicates that schools may make more of a difference to students’ knowledge compared to their competencies regarding citizenship outcomes. However, these results also show that, in general, schools can influence students’ citizenship competencies.

Overall, it could be stated that the findings reported above are only partly in line with functional community theory (Coleman & Hoffer, 1987) and previous findings. As we took on a rather explorative approach, we did not formulate clear-cut hypotheses prior to the analyses on the relationship between school composition and each of the social tasks. The main point of consideration was the differential role that homogeneity, consistency, and heterogeneity within contexts might play for human learning and development (Cochran, 1990; Coleman & Hoffer, 1987; Dijkstra et al., 2015). Similar to Dijkstra et al. (2015), we expected that more heterogeneous environments could be specifically favorable for learning about issues that concern contrasts between people. This expectation was only partly confirmed based on our findings.

As stated previously, to our knowledge, the study by Dijkstra et al. (2015) is the only study that investigated the relationship between school diversity and citizenship competencies in a comprehensive manner. In their study, a positive relationship was demonstrated between degree of school diversity and citizenship knowledge and reflection. As we used the same questionnaire as Dijkstra et al. (2015), but with a different operationalization for citizenship outcomes, it is somewhat difficult to compare the results. However, the most striking divergence is the fact that we found either a negative or non-effect of diversity on Knowledge depending on the social task, while Dijkstra et al. (2015) reported a positive effect on knowledge overall. It should be noted, however, that Dijkstra et al. (2015) used the CCQ for the measurement of knowledge while we used newly constructed knowledge items. Although both studies provide valuable insights, to further expand our knowledge on the association between school ethnic diversity and citizenship outcomes, more research in this area is warranted.

Taken together, our examination of the relationship between school composition and citizenship competencies yields some interesting findings, especially within the broader context of Dutch society. The Netherlands distinguishes itself with its highly diverse population and its strong international orientation. Currently, the country is home to over 200 different nationalities and is one of world’s largest exporters. Additionally, immigration in the Netherlands is ever-growing, bolstering the development of “super-diversity” (Vertovec, 2007), which indicates that the number of different ethnic groups in society is increasing and that the characteristics within different ethnic groups are also becoming more heterogeneous.

Against this background, the Netherlands contains fertile grounds for fostering citizenship competencies that are addressed in Critical Cosmopolitan Theory (Byker, 2013) and the social tasks approach (Ten Dam et al., 2011). These include the development of awareness and knowledge regarding cultural differences, interconnectedness, handling different perspectives, being concerned with global issues, and social (in)justice (Byker & Marquardt, 2016; Lilley et al., 2015). The phenomenon of super-diversity (Vertovec, 2007) also warrants a more urgent development of such competencies. However, at the same time, in the past decades, multiculturalism policies have been slowly terminated in the Netherlands (Banting & Kymlicka, 2013), and as is common in Europe, adverse attitudes concerning diversity are increasing.

The diversity in society is reflected in Dutch schools to different degrees, varying from schools with almost solely native Dutch students to schools that cater to students with various backgrounds. Our study has shown that the make-up of the student body results in mixed outcomes concerning citizenship competencies in Dutch schools. The effects of the school composition are either positive, negative, or non-existent. In a way, this finding resembles the social complexities that exist in society. Therefore, it can be stated that the findings are not surprising, but they are somewhat alarming. Although diverse schools demonstrate to be a beneficial environment for many of the social tasks, the findings conversely indicate that students in schools with predominantly native Dutch students evaluate themselves to be less competent citizens, which is worrisome in a societal climate in which one must more than ever rely on competencies to navigate society adequately. In addition, while being more competent, students in diverse schools have less knowledge on some of the social tasks that were investigated. In sum, the findings indicate that work remains to be done in promoting citizenship competencies of young people.

Regarding teacher support, positive associations were found with all social tasks and for both competence and knowledge. These findings match our expectation that students who perceive their teachers to be caring and fair (Ryan & Patrick, 2001) will learn more in the citizenship domain. The results additionally correspond to theoretical assumptions in the area of learning in general and the citizenship-related domain in particular (Cook, 1985; Flanagan, 2013; Flanagan et al., 2007; Wang & Degol, 2016). Finally, our findings are consistent with previous research on the relationship between teacher-student relationships and citizenship-related outcomes(Isac et al., 2014; Rutkowski et al., 2014; Wanders et al., 2020a, 2020b).

Interestingly, no moderation effect of teacher support was found, except for one outcome. A significantly negative interaction effect was demonstrated between school composition and teacher support in relation to acting in a socially responsible manner. This finding indicates that the more students feel supported by their teachers, the less they profit from being in a school that is highly diverse compared to students who perceive less teacher support. Or interpreted conversely, the less diversity in schools, the more positive the relationship becomes between teacher support and competence in acting socially responsible. The question remains whether the absence of a moderation effect for the other social tasks indicates an actual non-existing interaction. It could be the case that we would have been able to detect such a significant effect with a larger sample size and with more schools that are at the high end of degree of diversity.

Some limitations of our study should be considered before turning to the implications of our findings. First, the results are based on correlational data, which only enables the observation of a relationship between variables without any indication of causation. Future studies should take on a longitudinal approach to enhance insights on causal relationships and the developmental trajectory of citizenship competencies. Second, we used student self-reports for the measurement of both teacher support and citizenship competencies. In future studies, for the assessment of citizenship, perceptions of other informants, such as teachers, parents, and peers should be included (Geijsel et al., 2012). Such an approach could further broaden our understanding of young people’s citizenship competencies. Moreover, our findings reveal the importance of the amount of perceived teacher support. Follow-up research could additionally focus on teachers’ views of the support they provide, along with qualitative observational data. Lastly, our research is fully embedded in the Dutch context. For international insights, our research should be repeated with large samples in other countries.

In conclusion, our results imply that educational policy, schools, and teachers should pay attention to the characteristics of the student body and the degree of teacher support in their attempts to promote students’ citizenship competencies. Building strong teacher-student relationships appears meaningful for students’ capacities to carry out the social tasks that reflect their daily citizenship. As suggested by Wanders et al. (2020a), one way to improve students’ perception of teacher support is by allowing students to evaluate teachers. Prior research has shown that in schools that use student evaluations of teachers, teacher-student relationships are also better (Barile et al., 2012).

When it comes to preparing students for societal life, teacher education programs play a significant role (Byker & Marquardt, 2016). Research shows that in teacher training, preparation for handling diversity (of perspectives) is only limitedly addressed (Severiens et al., 2014). Policymakers should incorporate this overarching topic structurally in teacher education programs, as it is important that teachers’ “eyes are opened to the world” (Byker & Marquardt, 2016, p. 40) in the first place, in order to subsequently educate students who are engaged with society.

Lastly, our findings imply that schools and teachers should devote more attention to the school as a practice ground. At the time of writing, the Netherlands has passed a new bill concerning citizenship education (Dutch Ministry of Education, Culture and Science, 2019). Schools in the Netherlands are still offered a great amount of freedom in organizing their education the way they deem suitable, based on the constitutional right of freedom. However, in the past few years, this freedom has also resulted in the need for more guidance in the implementation of citizenship education. The new bill states that schools ought to develop and explicitly communicate their vision and policy regarding citizenship education. Interestingly, the new bill explicitly refers to the task of schools to function as a practice ground. Given the new bill, school administrators are obliged to ensure a school culture in which students can actively practice with the basic values of the democratic constitutional state and human rights. As our research shows the important role of school composition, we call on schools and teachers to critically evaluate the possibilities and challenges that their particular practice ground presents in relation to the student population. Our findings illuminate the potential of schools as mini-societies and practice grounds. Therefore, schools and teachers should take on the opportunity to contribute to students’ citizenship competencies and knowledge, considering the characteristics of the student population.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Dutch Organization for Scientific Research (NWO) under Grant 411-12-035.

Notes

1. For the current study we used correlational data, therefore no causal inferences can be made. However, we use statements such as “effect” whenever it contributes to the readability of the article.

2. To this end, a stratified sample was used based on the offered educational tracks of the schools (a division was made between pre-vocational, general, and mixed school tracks). For the sample, 100 schools were drawn from a list of all Dutch secondary education schools that include 9^th grade classes. If the first approached school did not confirm participation, a first or second replacement school was approached.

3. Although knowledge is also considered a competence, as stated, it differs in nature from the other components. For the sake of readability, we will therefore refer to the three combined components and knowledge as “Competence” and “Knowledge,” respectively. Thus, this only indicates a distinction of assigned labels.

4. Calculation of the index is as follows: (proportion ethnic group 1)² + (proportion ethnic group 2)² + · · · + (proportion ethnic group n)². The index developed by Putnam (2007) represents the degree of homogeneity. To obtain the degree of heterogeneity, we subtracted the resulting index from 1.

Appendix A:

Reliability coefficients, mean scores, and standard deviations of the two components of the social tasks (CCQ)

Table

	Number of items	Cronbach’s α	Mean score	Standard deviation
Acting democratically Knowledge ASR	77 18	.94 .83	.97 2.67	1.20 .38
Acting in a socially responsible manner Knowledge ASR	23 17	.85 .83	.94 2.59	1.30 .39
Dealing with conflicts Knowledge ASR	24 19	.86 .90	.86 2.77	1.28 .43
Dealing with differences Knowledge ASR	39 18	.90 .87	.79 2.44	1.26 .45

Appendix B:

Item examples of the CCQ and newly constructed knowledge items

Table

Component and social task	Introduction phrase	Item example	Response options (correct answer in bold)
Knowledge
Acting democratically	Instruction to choose the best option	“Laws..”	a) Are rules about what is and what is not allowed in a country b) Are made by the Senate c) Are only valid if all residents of a country agree to it
Acting in a socially responsible manner		“When you go on vacation it is best for the environment if you travel by..”	a) Car, then you can determine yourself which route to take b) plane, as the journey is shorter compared to other means of transport c) train, as it uses less energy compared to other means of transport
Dealing with conflicts		“Anouk and Lilian are having a fight in class. Babette is a mediator. What does Babette do then? Babette… (choose the best answer)	a) Tries to help find a solution b) Sees who is to blame in the fight c) Tells Anouk and Lillian what to do
Dealing with differences		“How is it that the Netherlands has become a multicultural society? That is because of…”	a) Segregation b) Migration c) Assimilation
Attitude
Acting democratically	How well does this statement apply to you?	“People must listen carefully to each other, even if they differ in opinion”	4-point Likert scale:
Acting in a socially responsible manner		“If I have had a picnic with friends in the park, it is normal for us to clean up the mess”	1) does not apply at all to me 2) does not apply much to me 3) applies a fair amount to me 4) applies completely to me
Dealing with Conflicts		“When I have a fight, I try to take the other person seriously”
Dealing with Differences		“I am curious about how people in other countries live”
Skills
Acting democratically	How good are you at..?	“Standing up for your opinion”	4-point Likert scale: 1) not good at all 2) not very good 3) pretty good 4) very good
Acting in a socially responsible manner		“Understanding what the other person feels”
Dealing with conflicts		“Coming up with a solution that everyone is satisfied with”
Dealing with differences		“Behaving normally in an unknown environment”
Reflection
Acting democratically	How often do you think about...?	“Whether pupils are listened to at your school”	4-point Likert scale: 1) (almost) never 2) only occasionally 3) fairly frequently 4) frequently
Acting in a socially responsible manner		“Why some pupils bully”
Dealing with conflicts		“What you did wrong yourself, after a fight”
Dealing with differences		“Why people of different origins do not interact much”

Appendix C:

Effect sizes of the significant primary independent variables

Table

	Acting democratically Competence	Acting democratically Knowledge	Acting socially responsible Competence	Acting socially responsible Knowledge	Dealing with conflicts Competence	Dealing with conflicts Knowledge	Dealing with differences Competence	Dealing with differences Knowledge
Degree of diversity	0.28		0.29	−0.39		−0.34	0.47
Teacher support	0.32	0.19	0.35	0.22	0.43	0.20	0.22	0.20
Degree of diversity x teacher support			−0.32