Exclusion and Antisystem Attitudes: The Impact of Perceived Discrimination in Attitudes Towards Democracy and the Willingness to Use Violence among Adolescents in Brussels

: Perceived discrimination, the perception of systematic exclusion due to background characteristics, has been studied extensively in general. The political consequences of this perception remain underexplored for adolescents. Discrimination may engender a rejection of common political values such as the support for democratic politics. Using the data of 1789 pupils with an average age of 16 years (grade 10) from 24 schools in Brussels, we focus on the consequences of perceived discrimination in attitudes towards violence, as well as on a rejection of representative democracy. The outcomes of a multilevel analysis suggest that high levels of self-reported perceived discrimination are significantly associated with an anti-democratic attitude (rejection of the current form of representative democracy) and the willingness to use violence. In a context in which 75% of pupils have a non-native background, these findings reveal the challenges for future forms of civic education.


Introduction
A growing number of inhabitants of European cities have an immigrant background (Schaeffer 2013). OECD (2012) states that "Today, foreign-born nationals constitute between 10 and 15 per cent of the population in Western European Countries" (Schaeffer 2013, p. 1). As a consequence, contemporary urban school contexts in these superdiverse cities are characterized by high degrees of ethnic diversity. Most of these new citizens do not see themselves automatically as full-fledged citizens of these societies. Using the European Social Survey (ESS), André and Dronkers (2017) found that in the 27 EU member states, the mean percentage of perceived discrimination is higher for immigrants (11.2%) than for natives (1.7%). This feeling of "being discriminated against" might affect the degree of social cohesion in these societies. We might also expect political consequences of perceived discrimination.
Research on perceived discrimination conceptualizes this phenomenon as a potential factor related to the perceived illegitimacy of authorities and violent radicalization (Doosje et al. 2013). In the political domain, research has documented the impact of perceived discrimination on the level of satisfaction with democracy (Ekman and Linde 2003; Ruiz-Rufino 2013) or on levels of political trust (Abrajano and Alvarez 2010). However, research on the impact of perceived discrimination, specifically in the context of school and daily life, is scarce.
We consider violence and a rejection of representative democracy as politically relevant attitudes that provide insight into the willingness to support (or reject) the prevailing democratic political system. Consequently, in the case of rejection, an anti-system attitude might affect the foundation of representative democracies. Therefore, measuring these two attitudes jointly might help us to better understand anti-system attitudes. Super-diverse cities in which the majority of the population has an immigrant background form natural laboratories to investigate discrimination and its consequences. To the best of our knowledge, this paper is the first to analyze superdiverse contexts related to perceived discrimination and attitudes towards the rejection of democracy and the willingness to use violence among disadvantaged youth. Since the rate of perceived discrimination is higher among immigrants than among natives (André and Dronkers 2017) and perceived discrimination can impact political attitudes (Takyar 2019;Oskooii 2018;Sanders et al. 2014), we might expect that, within these specific superdiverse contexts, perceived discrimination will have detrimental effects on the attitudes towards and expectations of democracy and society, especially for young people from a 'non-native' background With 62% of its inhabitants having their origins outside of Belgium, the Brussels Capital Region is, according to the World Migration Report, the second most diverse city in the world (Lee 2015). In 2017, 71.9% of the inhabitants of the Brussels Capital Region did not initially possess a Belgian nationality on the day they were born, or had at least one parent for whom the Belgian nationality was not their first nationality (Statistiek Vlaanderen 2019, p. 7). In addition, the Brussels Capital Region has two official languages (French and Dutch), but is in fact a multilingual context where English has become the second most common language in use (Janssens 2018). Furthermore, after three decades of urban flight, the Capital Region is the fastest growing and 'youngest' region of Belgium. Brussels has seen a 'youth bulge' with almost one-third of its population being younger than 25 (Sacco et al. 2016). This makes the city an interesting research site to investigate superdiverse societies (Neudt and Maly 2010). Sacco et al. (2016) show, in one of the few summarizing articles regarding Brussels adolescents, that different scholars point out that ethnical, social and school mechanisms disadvantage this 'youth bulge'. Pupils with an immigrant background are represented disproportionally highly in nonacademic (i.e., technical and vocational) pathways (Jacobs and Rea 2007). The French and Flemish school systems also disadvantage pupils from lower social strata since there is a strict segregation between "good" and "bad" schools. The former are schools with mostly pupils from elite and white backgrounds, the latter are schools with a high rate of pupils with a low socio-economic and immigrant background (Janssens et al. 2009). The current societal situation of Brussels adolescents is alarming since they "have a high percentage of school drop, low grades at school and high rates of unemployment" (Sacco et al. 2016, p. 6;Pitts and Porteous 2005;Pitts and Porteous 2006) In addition, the presence of perceived discrimination could prove to be deleterious to schools' efforts to promote social cohesion (Putnam 2007;Laurence 2011;Portes and Vickstrom 2011). To be able to support and empower disadvantaged youth in this specific context, it is necessary to investigate their attitudesspecifically, how perceived discrimination (in the context of school and daily life) influences attitudes towards the willingness to use violence and the rejection of representative democracy. These two attitudes form the fundamental basis to measure anti-system attitudes. Initially this concept was coined by Sartori in the 1960s and 1970s to analyze party systems (Capoccia 2002). Nevertheless, we want to use this concept in an empirical manner and operationalize it to measure antisystem attitudes among adolescents and to what extent they are willing to use violence and reject representative democracy.
Using data from a survey 1 of grade 10 students (N = 1789-average age 16 years) from the Brussels Capital Region, we aim to contribute to our understanding of the impact of perceived discrimination on broader attitudes. How does perceived discrimination impact the attitudes of adolescents vis-à-vis the rejection of representative democracy and the willingness to use violence?
In the following paragraphs, we review the literature regarding perceived discrimination, 1 Democratic Empowerment of Brussels Education, Students and Teachers (DEBEST).
attitudes towards the rejection of representative democracy and the willingness to use violence, and how the latter two are related or influenced by the former. Secondly, we analyze the impact of perceived discrimination on attitudes towards democracy and the willingness to use violence, using recent cross-sectional survey data on adolescents. We conclude with suggestions for further research and a few key limitations.

Perceived Discrimination
Originally the act of noting differences, discrimination now denotes differentiation between people on grounds such as "gender, color, sexuality, disability or class" (Rai 2018, p. 163). Perceived discrimination is known as one of the main early conditions of (immigrant) children's political integration since it crystalizes a context of exclusion on a "group-level" (Esser 2015). Therefore, perceived discrimination is damaging for the political empowerment of adolescents with an immigrant background and might cause alienation from representative democracy in superdiverse cities and countries. Jacobs and Rea (2007) concluded that the perception of stigmatization among ethnic minorities is significant among adolescents of Moroccan and Turkish origin in Brussels. In addition, adolescents in Brussels experience stigmatization and racism in the form of verbal insults by different perpetuators, such as staff on public transport, bouncers, security staff, police staff and teachers (Jacobs and Rea 2007). In particular, the insults by teachers (17.7%), can be very problematic, as youngsters are subjected to compulsory schooling until the age of 18, and teachers exert a position of power and influence (Jacobs and Rea 2007).
The literature on discrimination distinguishes between perceived personal or egocentric discrimination (Pascoe and Richman 2009;Sanders et al. 2014) and group or sociotropic discrimination (Taylor et al. 1990;Sanders et al. 2014). The first is discrimination that is individually experienced by a person. The second is "the sense that members of one's own ethnic group suffer discrimination, regardless of one's own personal experiences" (Sanders et al. 2014, p. 125). Other authors differentiate between perceived political (institutional) discrimination and perceived societal (interpersonal) discrimination. The first refers to discrimination by institutions, whereas the latter defines discrimination between individuals and results in more problematic forms of political nonparticipation (Oskooii 2018).
Since there are several studies that show the crucial role of adolescents' perceived discrimination at school and in daily life in relation to psychological distress (Pascoe and Richman 2009;Priest et al. 2013;Sanchez et al. 2015;Schmitt et al. 2014), we chose this as an important parameter to understand the experiences of adolescents. Our aim is to enquire how perceived discrimination relates to attitudes towards the willingness to use violence and the rejection of democracy. Moreover, we are interested in perceived discrimination and are not seeking to gauge this perception via an unbiased or objective scale.
Our focus is on perceived societal discrimination and perceived institutional discrimination, mainly in the school context; we will try to empirically ascertain whether perceived discrimination enforces attitudes towards violence and alienates adolescents from representative democracy. We use the concept of Capoccia (2002), which contains the following two fundamental components: "More generally, the label anti-system has been used for a party or a group with non-democratic ideals" (Daalder 1966a(Daalder , 1966bBudge and Herman 1978;von Beyme 1985;Ferraresi 1988) or whose supporters or members engage in unconventional, illegal or violent behavior" (Zimmermann 1989;Capoccia 2002, p. 12). The anti-system concept appears to be "stretching" (Sartori 1970;Capoccia 2002, p. 10); however, the basis of the concept is solid enough to enable its use as a parameter for adolescents' attitude towards the current form of representative democratic system and the willingness to use violence. It is important to emphasize that being prone to violence is not the same as having an antisystem attitude. We underline that an antisystem attitude can solely be measured if the two scales, antidemocracy and the willingness to use violence, are measured jointly.

Perceived Discrimination and Attitudes towards Violence
The association between perceived discrimination and violent attitudes is important to investigate in established democracies with high ethnic diversity, such as in several European cities. First, this is because these locations have become defined by ethnic diversity and ethnic minorities tend to have an unequal position in society (Alanya et al. 2015;van Bergen et al. 2015) and, second, this is also because violent attitudes as a function of perceived discrimination may reflect nonconformity with the way the democratic system is functioning. The willingness to resort to violence for achieving goals may be a signal of citizens questioning the legitimacy or effectiveness of the institutional channels or authorities in representing their interests (Doosje et al. 2013;Schwarzmantel 2010).
We argue, moreover, that the relevance of investigating people's willingness to use violence is timely in the ongoing debate about the legitimacy crisis facing consolidated democracies (Van Ham et al. 2017;van Beek 2018. Accordingly, the resort to violence to resolve differences in these societies may reflect the failure of consolidated democracies to provide an inclusive political community that accommodates a plurality of beliefs and values (Schwarzmantel 2010).
In the European context, recent studies have concentrated on examining perceived discrimination as a predictor of violent preferences.
For example, a study among young Dutch Muslims assessed the relationship between perceptions of authorities as illegitimate and violent attitudes. This study is relevant because it argues that the perceived illegitimacy of authorities may be related to previous experiences of perceived discrimination. The study's results conclude that when people perceive the authorities as illegitimate, they are more likely to hold favorable attitudes towards violence from other in-group members. In turn, this attitude was a predictor of their own intentions to use violence (Doosje et al. 2013).
Studies conducted in Belgium among young people show more mixed evidence on the conditions that are relevant to the discrimination-violence nexus. A study conducted among pupils in French-speaking secondary schools revealed that perceived personal discrimination associates positively with males' "non-conventional/illegal political engagement", such as burning flags or writing graffiti with a political message. In contrast, group discrimination was found to be insignificant among all genders (Gavray et al. 2012). However, a study assessing different predictors for Flemish youth involvement in politically motivated violence towards property and persons found group discrimination to have a strong positive association with self-reported political vandalism, while perceived personal discrimination tended to be less relevant (De Waele and Pauwels 2014).
Perceptions of personal or group discrimination are relevant not only with respect to violent attitudes; the literature suggests that the frequency and settings where discrimination takes place are also relevant. Along these lines, a study of Dutch-speaking secondary schools in Belgium found that discrimination based on the grounds of politics or language rather than religion was associated with violent radicalization. In addition, young people who had experiences of discrimination while interacting with the justice system were more likely to endorse violent extremism. Reporting more reasons for being discriminated against and more settings in which discrimination was experienced was also a meaningful association (Frounfelker et al. 2019).
In accordance with previous studies, we hypothesize: Hypothesis 1 (H1). Adolescents with more experiences of perceived discrimination are more likely to have positive attitudes towards the use of violence than adolescents with less experiences of perceived discrimination.
Overall, previous research suggests that there may be a positive association between discrimination and violent tendencies. More research is needed to understand the conditions that foster this association and whether it can be generalized to different minorities within a society. Unfortunately, these studies tend to emphasize violent behavior but overlook how discrimination is linked to attitudes towards violence. In addition, they do not reflect on what the very act of resorting to violence may reveal about the (dys)functioning of representative democracies. Our study attempts to enquire precisely how this association is unveiled in a superdiverse and multilingual city such as Brussels. We specifically advance, in two senses, the understanding of the nexus between discrimination and violent attitudes.
The literature highlights the importance of 'unpacking' discrimination (Frounfelker et al. 2019) to assess its consequences on the endorsement of violence. This implies the assessment not only of whether young people have experienced perceived discrimination, but also of the settings where discrimination is experienced and its frequency (Frounfelker et al. 2019;Rousseau et al. 2018;Alanya et al. 20115 . It bears mentioning that those settings need to reflect the common experiences of young pupils (Pachter et al. 2010) to reveal a meaningful association with violent attitudes. Thus far, validated measures specifically designed for adolescents when evaluating this association have not been applied in the literature. This research fills that gap.

Perceived Discrimination and Attitudes towards Democracy
Empirical research regarding the relationship between perceived discrimination and attitudes towards democracy-specifically attitudes towards the rejection of democracy-is scarce in the literature. Most studies document the relationship between perceived discrimination and voting behavior or trust in official institutions. They show a negative relationship between perceived discrimination and voting (Schildkraut 2005), trust in government (Abrajano and Alvarez 2010;Maxwell 2009;Michelson 2003) or a positive relationship between perceived discrimination and nonelectoral political activities (Heath et al. 2013). In 2014, the data from the European Social Survey (ESS) showed a negative relationship between perceived discrimination and the level of satisfaction with democracy among second and later generations of ethnic minority migrants (Rood 2018).
These findings show that having a low satisfaction with regard to democracy or feeling disconnected from the country in which one lives are attributable to a reduced sense of national belonging or social exclusion, and that this might enforce the rejection of democracy.
In contrast to research regarding the satisfaction with democracy, 'that measures whether democracy is functioning as it should' (Rood 2018, p. 4), our research aims to measure the impact of perceived discrimination on antidemocracy or to what extent young people are rejecting today's representative democracy (Elchardus and Tresignie 2002;Kavadias 2004).
One of the few studies that have close similarities to our research was conducted in Quebec, Canada (Bilodeau 2017). The researchers observed that foreign-born and native-born minorities have a negative relationship between perceived discrimination and a low satisfaction with democracy (Bilodeau 2017). On this basis, we hypothesize that: Hypothesis 2 (H2). Adolescents with more experiences of perceived discrimination are more likely to have a negative attitude towards democracy than adolescents with less experiences of perceived discrimination.

Design
With 62% of its inhabitants having an origin outside of Belgium, the Brussels Capital Region was, in 2015, the second most diverse city in the world (Lee 2015). In 2017, 71.9% of its inhabitants either had no Belgian nationality or at least one parent without Belgian nationality on the day they were born (Statistiek Vlaanderen 2019).
The high degree of diversity alongside social diversity makes Brussels a superdiverse city (Vertovec 2007). Moreover, the Brussels Capital Region has two official linguistic communities, as well as two separate educational systems (French and Flemish). As a consequence, the 'Capital of Europe' is an archetypical superdiverse metropole, but also an exceptional case because of the bilingual institutional construction.
The respondents of our survey, grade 10 pupils, are on average 16 years old. It is around this crucial age that adolescents form their civic and political attitudes (Erikson 1994). In all likelihood, their basic personality will, in most cases, probably not change significantly during this stage of their life (Inglehart 1990;Jennings and Niemi 1981).

Sample Selection
We used survey data collected during the school year of 2018-2019, in Dutch-speaking (Flemish) and French-speaking secondary schools in the Brussels Capital Region. Education in Belgium is compulsory from the age of 6 until the age of 18. Primary schooling begins from the age of 6 and continues until the age of 12, and secondary education continues until the age of 18. Pupils are tracked according to academic capabilities into general technical/artistic or vocational tracks after the second year of secondary education (De Groof and Franck 2013).
For the survey, we selected 24 secondary schools via stratified random sampling, with language community, location, governance structure and tracks as the strata. First, we selected schools from both official language communities in the Brussels Capital Region: Dutch and French speaking. The sampling for the survey was drawn on the basis of all secondary schools of the Flemish community in the 19 municipalities of the Brussels Capital Region. For each sampled Dutch-speaking secondary school, the geographically nearest secondary schools of the French-speaking community were selected. As Brussels has a clear socio-geographic division between rich and poor areas, we also used school location as a criterion (Sacco et al. 2016). Thirdly, as Belgium has both public schools and statesponsored private (mostly Catholic) schools, we used the type of governance as a third criterion. The fourth criterion was based on the results of educational tracking: some schools offer only a general (academic) curriculum, while others offer technical or vocational education. This design of the sampling allowed us to guarantee the presence of different profiles (state-sponsored private/public schools, educational tracks) for each language community and to select schools from the same neighborhoods. In each secondary school, the pupils in their fourth year (grade 10) were asked to fill in the questionnaires in class (after being informed and having given their consent). To correct sampling biases, we computed post-stratification weights according to gender, language, the governance structure of the school (public/private) and tracks (academic/non-academic) on the basis of population data (Little 1993).
Since we dealt with clustered school data, we used multilevel models. The data were analyzed with SPSS 26 for data description and preparation. The multilevel analysis was performed in R (using the lme4-package). We controlled for possible confounding variables. First, we measured the impact of perceived discrimination on the willingness to use violence and, subsequently, on anti-democracy attitudes. Second, we added the control variables.

Operationalizations
To measure to what extent adolescents had an antisystem attitude, we used two outcome variables. One scale measures the willingness to use violence and one scale gauges the attitude towards current (representative) democracy. The first outcome variable was made by Doosje et al.
(2013) using the following Likert items: "I am prepared to use violence against other people in order to achieve something I consider very important", "I am prepared to disturb the orderliness in order to achieve something I consider very important"," I am prepared to destroy things in order to achieve something I consider very important". The answer categories were: "I totally agree", "I agree", "Neither agree nor disagree", "I do not agree", "I totally do not agree".
The second outcome variable, a measure for "anti-democracy", was created by Elchardus and Tresignie (2002). This variable measures whether the respondent rejects representative democracy, using the statements: "The so-called experts and specialists know nothing", "Democracy is just a veil for the power of the rich", "Without political parties our country would be much better off". The answer categories were once again, "I totally agree", "I agree", "Neither agree nor disagree", "I do not agree", or "I totally do not agree".
To measure perceived discrimination, we used a reduced form of a scale proposed by Pachter et al. (2010) (Table 1), using 8 of their 23 proposed items. Since adolescents spend a great deal of their time in school, we selected four items that refer specifically to situations that are related to pupils, teachers and the classroom. The other four items refer to circumstances that are strongly recognizable in the daily life of adolescents.
Pupils were asked, "Have you had the following experiences?" with the following possible responses: "Followed by security guards at stores", "Treated badly/unfairly by teacher", "Poor/slow service at restaurant", "Got grades you didn't deserve", "People hold bags tight when you walk by", "Someone was afraid of you", "Teachers assume you are not intelligent", "Treated badly/unfairly by teacher". The answer categories were: "never" (=1), "once" (=2), "several times" (=3), "regularly" (=4). Table 1 presents the frequency distribution of the question measuring perceived discrimination (Cronbach's alpha: 0.83). Items showed that, on average, perceived discrimination in schools is more prevalent than perceived discrimination in other settings, particularly for the following experiences: "accused of something you did not do at school", "got grades you didn't deserve", "treated badly/unfairly by a teacher". Moreover, a Principal Component Analysis (PCA) of the independent variable "perceived discrimination" shows that there are clearly two subcomponents or subdimensions: perceived discrimination at school and perceived discrimination in a wider environment.
Lastly, pupils who identify themselves as Muslim and pupils who have a Moroccan ethnic background score highest on the perception of discrimination. Likewise, Table 2 presents the frequency of the question measuring the willingness to use violence (Cronbach's alpha: 0.83); Table  3 presents the items measuring antidemocracy (Cronbach's alpha: 0.63). For further analyses, we use standardised variables (z-scores).  As control variables, we included gender, social-economic status, origin, religious identification and school track (Table 4). Gender is coded as 0 (=male) and 1 (=female). Socio economic status (SES) was operationalized as the highest attained level of education of the father and mother (or the person who has the role of the father or the mother in the household). We recoded the original variables in three categories: "low" (from no education, to lower secondary school education (15 years) at most), "medium" (secondary school education), "high" (higher education). The third control variable is the ethnic origin of the pupil. We categorized their countries of origin into seven groups: "Belgium", "Morocco", "Turkey", "Africa", "EU15", "Europe" and "other". As perceived discrimination was also observed to be dependent on the respondent's religion, we asked youngsters to choose the religion with which they identify out of a list of eleven options (the twelfth option was "I'm not interested in anything that involves religion"). We recoded this variable into four categories, distinguishing "Muslims" from "Christians", "non-believers", and a residual category of "other". Since separating pupils in terms of tracking might affect positive attitudes towards democracy (Kavadias et al. 2017), it is important to include school track as a fifth control variable, differentiating between "general", and "technical/artistic/vocational" tracks.

Perceived Discrimination and the Willingness to Use Violence
We estimated six multilevel regression models in order to explore and explain the dispersion of perceived discrimination and the willingness to use violence among subgroups (Table 5). In the second model, only gender is included. In the third and the fourth, we added the education background of the parents and the origin of the grandparents, respectively. The fifth model contains religion and the sixth model contains school track. This finding shows that the initial bivariate relationship between perceived discrimination and the willingness to use violence was already strong and significant after controlling for different background characteristics, and that there remained a positive and significant relationship between perceived discrimination and the willingness to use violence (Model 1, 2, 3, 4, 5 and 6). These findings are in accordance with our first hypothesis. Moreover, the impact of perceived discrimination is strong, with a beta of 0.32 after all the controls. The relationship between perceived discrimination and the willingness to use violence shows that the more that young people perceive discrimination, the more they are willing to use violence against other people, disturb the order and destroy things.
Secondly, the impact on the use of violence is significantly lower amongst girls. This is in line with the findings of Rousseau et al. (2018). Furthermore, pupils with more highly educated parents have a significantly higher tendency to use violence. Pupils of Belgian origin score lower on the use of violence than the reference group. In addition, pupils identifying themselves as Christian, atheist and pupils from other religions or philosophies score significantly higher than Muslim youngsters regarding the use of violence.

Perceived Discrimination and Anti-Democracy
We followed the same steps for our measure for "anti-democratic attitude" (Table 6). Again, we see an association between perceived discrimination and anti-democracy attitudes (beta: 0.22). Controlling for different social background characteristics does not alter this relationship, which leads us to accept the second hypothesis (Model 1, 2, 3, 4, 5 and 6). Table 6 shows that, after controlling, the beta remains stable at 0.22.
Again, we find a very strong and significant effect on antidemocracy with gender, but this time the female respondents tend, on average, to exhibit a higher degree of rejection of representative democracy than the boys. Furthermore, two ethnicities show significantly less association with antidemocracy attitudes. Pupils of Turkish origin score significantly lower for anti-democracy attitudes and pupils of Belgian origin show a significantly low association with anti-democracy attitudes. Lastly, there is no association between religious self-identification and attitude towards representative democracy in the final models. Cell entries are unstandardised regression coefficients. Cell entries are unstandardised regression coefficients.

Discussion and Conclusions
An important cornerstone of any democratic political regime is the expression of people's interests through institutional channels. However, our study shows that when people do not feel represented by these institutions/authorities or when they are seen as inherently biased towards the benefit of certain groups due to feelings of personal discrimination, their legitimacy (Doosje et al. 2013) and effectiveness to channel people's interests can be called into question. In this scenario, alternative channels of expression may be justified, including non-conventional forms of political participation and the use of violence. In this sense, it is not surprising that other studies have found that, for people who perceive discrimination against them, non-civic activities tend to be significantly related to other types of conventional civic participation (Gavray et al. 2012). Therefore, people with experiences of perceived discrimination are not necessarily deprived of collective action. As the previous literature has concluded, perceived discrimination can trigger collective mobilization. However, people that feel discriminated against would need a strong ethnic identity to be mobilized (Stronge et al. 2016). Moreover, given that our results show that people with perceptions of being discriminated against were found to score higher on the willingness to use violence, we would expect that this collective action would not only be expressed through conventional and legal channels but could also be expressed through a violent outlet.
In this paper, we attempted to achieve an empirical grasp on the relationship between perceived discrimination, the willingness to use violence and the rejection of representative democracy ("antidemocracy") among adolescents. In general, the higher that adolescents score for perceived discrimination, the more they reject democracy and are willing to use violence.
Seemingly, these attitudes exteriorize and can be considered as proxies for an anti-system attitude. Perceived discrimination alienates young people from representative democracy, pushing them into the margins of society where their anti-system attitude only festers.
It is worth noting that female pupils score significantly higher on anti-democratic attitudes compared to boys. Further research should clarify this difference. Furthermore, the willingness to use violence among the following pupils was remarkably significant: those from a country other than the six categories, those from a country in Europe other than the 15 core-countries, those identifying as atheists and those with another philosophy or religion (than Christian or Muslim). Further research should clarify these differences.
The implications of our findings are particularly revealing for the (dys)functioning of consolidated democracies. Our findings warn against the negative side effects of civic attitudes that perceived personal and perceived institutional discrimination may trigger in established democracies. These side effects are related to the endorsement of violent attitudes that are detrimental to the optimal functioning of a democratic regime. It is important to conduct further research to counter this perceived discrimination and its negative impact on democracy, regarding the determinants or predictors of perceived discrimination in the context of schools. Seemingly, there is a possible mismatch between the aspirations of, on the one hand, teachers, curricula, principals and school policies and, on the other hand, pupils regarding the idea of how to function as a pupil at school and which political system is acceptable. Although we cannot realistically expect civic education courses to 'fix' this mismatch, one could consider setting up a form of democratic dialogue between these pupils and their schools to bridge the disparities. This, however, necessitates more indepth research to understand, for example, what exactly happens at a school regarding items of perceived discrimination. How do pupils and teachers interact precisely in daily life when pupils find that they are "accused of something they did not do at school", "got grades they didn't deserve", or were "treated badly/unfairly by a teacher"? Why exactly do pupils reject representative democracy and why are they willing to use violence? In addition, we need more insights into the values of pupils and teachers concerning violence, democracy and the school system. This could clarify the complex dynamic between perceived discrimination, an antisystem attitude and the role of the educational system.
In brief, the impact of perceived discrimination on the willingness to use violence and the rejection of representative democracy should be investigated further to avoid political alienation (Durkheim [1897(Durkheim [ ] 1951Doosje et al. 2013; Hoskins and Janmaat 2019 and anti-system attitudes that could diminish social cohesion (Putnam 2007;Laurence 2011;Portes and Vickstrom 2011), which can foster social polarization (Esteban and Schneider 2008). We also see these violent attitudes as a warning regarding possible alienation from representative democracy, pushing people into the margins of society where their anti-system attitude only festers. This is something that young people who feel discriminated against may experience from the political system. These negative consequences add to the already vast evidence documenting the detrimental effects of perceived discrimination on other areas of young people's personal development such as wellbeing (Priest et al. 2013;Schmitt et al. 2014;Kauff et al. 2017;Benner et al. 2018;Giuliani et al. 2018).
The core of our study is that perceived discrimination has an important impact on the willingness to use violence and the rejection of representative democracy. These two variables manifest in parallel regarding perceived discrimination. Therefore, we conclude that perceived discrimination-specifically at school-is an important predictor.
Although we feel confident that our results offer a good starting point for further research, we are ready to acknowledge the limitations of our study. One limitation of our study is that we are not able to discern whether the attitudes towards violence are related to a political or religious motivation. We are only certain of a general willingness to use violence to reach perceived important goals in life. In contrast, we are able to bring more nuance to the different forms of discrimination by assessing different settings in which these experiences took place, namely that perceived discrimination scores highest in the educational context. Furthermore, our analysis points to a possible problematic relation between the current democratic and education context and the younger generations with superdiverse backgrounds. Our cross-sectional analysis is, however, not able to sketch out the precise dynamics of how this tension evolves. Another more technical limitation is that we used a weighting factor because of the low number of pupils in French-speaking schools with parents who had a low level of education, resulting in that cohort being underrepresented. We were also limited in our measurement of perceived discrimination as we only gauged personal perceived discrimination. The used measurement scales did not enable us to compare the difference between personal discrimination and group discrimination (Oskooii 2018;Sanders et al. 2014;Taylor et al. 1990). A last limitation is that we did not enquire as to the political "left" or "right" position of pupils. Since there is no research regarding perceived discrimination and political positioning, this should be investigated in other research. Although we did not explicitly approach other political variables, like, for example, authoritarianism, a lack of political efficacy or political cynicism, we want to highlight some preliminary findings. There are weak correlations between perceived discrimination and political knowledge and between perceived discrimination and authoritarianism. In contrast, there is a strong correlation between perceived discrimination and a lack of political efficacy. Funding: We received grants from the Universitaire Associatie Brussel, in support of the research program 'Democratic Empowerment of Brussels Education Students and Teachers'.

Conflicts of Interest:
The authors declare no conflict of interest. The founding sponsors had no role in the design of the study; nor in the collection, analyses, nor interpretation of data; nor in the writing of the manuscript nor in the decision to publish the results.