Measuring and explaining political tolerance among adolescents: insights from the International Civic and Citizenship Education Study 2016

Background Tolerance is a prerequisite for deliberative democracies. Therefore, fostering tolerance is an important task for educational systems in democracies. In the present study, the concepts of social and political tolerance were disentangled and applied to the measurement approaches of the International Civic and Citizenship Education Study 2016 (ICCS 2016). A single scale of political tolerance was proposed by identifying items on equal rights of women and men, different racial and ethnic groups and immigrants that capture the support of equal political rights. Previous research indicates different approaches that might be used to foster tolerance in schools. Methods This study used the rarely applied alignment method to identify a political tolerance scale that was sufficiently invariant in the 15 participating European educational systems. The alignment method identifies solutions with minimal amounts of non-invariance. Multilevel regression analyses are used to identify which aspects of schooling are predictive of political tolerance at the student and at the school level. Results The alignment analyses revealed a sufficiently invariant solution for the 15 education systems that could be used in regression analyses. While cognitive learning showed only a weak relationship with political tolerance, teacher fairness was relevant at student and school level. An open classroom climate was a positive predictor with relevant strength only at school level. Further multigroup analyses indicated that the results at the student level were stable between countries, while school level indicators varied more strongly between countries. Conclusions Researchers need to acknowledge the conceptual differences between social and political tolerance in their measurement approaches. The alignment method can be useful to create scales that are comparable between many different educational systems. While educational systems differ strongly in their school level variance of political tolerance, teachers can play an important role in fostering political tolerance at the student level.


Introduction
Fostering support for equal rights for different groups is an important topic for schools in diverse and democratic countries.Political tolerance, as the support of the right of everyone to participate in political processes, is important for collective decision making processes (Sullivan & Transue, 1999).Values, such as tolerance, are a cornerstone of democratic systems and an aspect of political support.Political support is a concept from research on political culture and has been described as a form of energy that citizens provide to the political system.This energy is needed for political systems to both persist and to evolve (Easton, 1965;Norris, 2017).Populist strategies and the dangers they pose to political support and democratic structures have recently received increased attention (Abs, 2021;Noack & Eckstein, 2023;Norris & Inglehart, 2019).Populism constructs opposition by a unified, homogenous people to a disloyal elite.While some researchers discuss the merits of populism for fostering change in a political system (Mouffe, 2021), populism generally polarises societies and undermines political support, especially the value of tolerance (Abs, 2021;Kinnvall & Svensson, 2022).While there is very little knowledge on the opportunities that schools have to combat populism (Noack & Eckstein, 2023), researchers do have the chance to investigate the role of schools in developing tolerance (Janmaat, 2022).As democratic processes rely on the mutual recognition of participants as equals (Habermas, 2003), measuring and fostering tolerance in young, emerging citizens is a critical task for policymakers, researchers, and practitioners.
Fostering tolerance is a goal for all education systems.In its 1995 Declaration of Principles on Tolerance, the United Nations Educational, Scientific and Cultural Organization defined tolerance as 'respect, acceptance and appreciation of the rich diversity of our world's cultures, our forms of expression and ways of being human' .The declaration demanded that states should adopt impartial political processes to protect groups from marginalisation ( § 2).Further, it stressed the relevance of tolerance at the individual and the community level ( § 3), and the value of education for tolerance ( § 4) (UNESCO, 1995).This normative document is substantiated by theoretical considerations from different disciplines, such as research on bullying, and civic and citizenship education.Recent findings in research on bullying have emphasised the relevance of prejudice in the process of bullying (Minton, 2014;Ziemes & Gutzwiller-Helfenfinger, 2019).Intolerant schemas such as sexism, homophobia, and racism can play a key part in the selection of bullying victims and the process of bullying itself (Goodboy et al., 2016;Minton, 2014).In research on civic and citizenship education, tolerance is regarded as an important prerequisite for deliberative processes (Sullivan & Transue, 1999).If members of a social group are not considered as deserving of equal rights, their perspectives in decision-making processes will not be deemed worthy of respect.Tolerance and the support of equal rights are seen as a core value for citizens (Sherrod & Lauckhardt, 2009) and their relevance is increasing relevance is increasing due to equally rising levels of diversity within classrooms (Isac et al., 2018).Tolerance is part of a network of attributes that students can develop as emerging citizens.
While many studies have investigated the predictors of tolerance (Beelmann & Heinemann, 2014;Pettigrew & Tropp, 2006), there is often little clarity concerning its very conceptualisation.This study explores different definitions of tolerance and applies them to the International Civic and Citizenship Education Study (ICCS).Based on the discussion of the literature, a conceptualisation of political tolerance will be proposed.Further, the rarely applied alignment method will be used to verify that political tolerance can be measured in adolescents in different countries.Finally, by using multilevel analyses, the study investigates student-and school-level characteristics that are important for the development of political tolerance in adolescents.

Conceptualising tolerance
The concept of tolerance has historically been and continues to be widely used in research, and therefore many definitions have arisen and are circulating.This multitude of approaches can lead to confusion among researchers and practitioners alike as to whether any two people refer to the same construct.To enable a shared understanding of tolerance, this section briefly describes the main conceptualisations and approaches for measuring tolerance and then positions the measurement method of the ICCS 2016 study within this framework.Based on the literature review, a distinction will be made between social and political tolerance.While social tolerance refers to interpersonal attitudes and behaviours concerning other groups, political tolerance refers to beliefs regarding how the state should organise the rights and responsibilities of different groups (Gibson, 2006;Viegas, 2007).
The discipline of psychology mostly focuses on the conceptualisation of social tolerance.Social tolerance can be understood as the opposite of prejudice and (internalised) negativity.Prejudice and internalised negativity refer to the negative affect and beliefs concerning groups that are marginalised by homophobia, racism, or sexism.These attitudes can be held by members of the dominant group and make discrimination and violent behaviour more likely (Schütz & Six, 1996;Wagner et al., 2008).However, these attitudes can also be internalised by the members of the marginalised group, and they are related to depression and other negative psychological outcomes (David et al., 2019;Meyer, 2003;Szymanski et al., 2009).Researchers of internalised negativity have argued that marginalised and normalised people alike internalise negative attitudes by living in a culture that is oppressive towards these discriminated groups (James et al., 2003).Moreover, researchers have discussed possible further differentiations of prejudice, for example between subtle and overt prejudice (Pettigrew & Meertens, 1995), implicit and explicit prejudice (Son Hing et al., 2008), as well as the distinction between the absence of prejudice and the presence of appreciation of differences (Hjerm et al., 2020).The different approaches share a common emphasis on individuals and their attitudes and behaviours towards different groups and their members.Measurement approaches have typically focused on asking participants how they feel about members of specific groups, how they would behave towards group members or, when using implicit association tests, what non-conscious associations they have with any specific group.
Tolerance has different connotations in the political sciences.When political scientists write about tolerance, they mostly refer to beliefs concerning the rights of different and opposing groups.These approaches are less interested in the interpersonal dimension and more interested in what people believe the rights of groups within a political system should be (Gibson, 2013;Kymlicka, 1992).Habermas (2003) used 'tolerance' strictly to denote a group's acceptance of the rights of other groups that are similar in power and that are not marginalised.In contrast, some researchers have referred to the support for rights of groups that are deemed as other (Kymlicka, 1992).This difference has been reflected in variations within approaches to measuring tolerance.While some researchers have tried to identify the social groups least liked by study participants, others have taken a more open approach and have explored broader concepts of political tolerance that include disliked groups in general (Gibson, 1992(Gibson, , 2013)).Maurissen et al. (2018) used the term 'political tolerance' when asking students to rate their acceptance of migrants without, however, asking them if they disliked migrants.This reflects the idea that marginalised groups are marginalised for the very reason that many people and the culture as a whole share internalised negative attitudes against them.If we accept this notion, asking participants if they dislike members of marginalised groups becomes less relevant.While there are different conceptualisations of political tolerance, they share a focus on attitudes towards the rights of other groups within a system.
Based on this review, the different approaches towards measuring social and political tolerance can be distinguished according to whether they focus on individuals' emotions and personal interactions, or individuals' attitudes towards state rules concerning the rights of different groups.While social tolerance seems to be more important for promoting positive relationships among individuals with different group memberships, political tolerance is more relevant for processes of democratic deliberation.In this paper, political tolerance is understood as the willingness to extend rights to disadvantaged groups to participate in political decision-making processes.That notwithstanding, social tolerance is related to political tolerance.The pervasiveness of internalised negative attitudes towards a multitude of groups renders a conceptualisation of political tolerance that focuses on tolerance of groups with an equal social standing less viable for research and in practice.

Tolerance in ICCS 2016
This section discusses how tolerance was assessed in the ICCS 2016 study.ICCS aimed to assess how well students are prepared for their roles as citizens (Schulz et al., 2017).Within the assessment framework of ICCS, attitudes towards equal rights belong to the civic principles that are 'the shared ethical foundations of civic societies' .The framework does not explicitly connect the support of equal rights to the concept of (political) tolerance, but it does link it to notions of equity and human rights (pp. 18-20).The next section will lay out to which degree the items can be used to measure aspects of political tolerance.
ICCS 2016 employed three scales concerning the support of equal rights as follows: first, equal rights of men and women; secondly, equal rights of racial and ethnic groups; and thirdly, the equal rights of immigrants.All scales included items concerning political and economic equality.The scales concerning the rights of racial and ethnic groups as well as the scale concerning the rights of immigrants additionally included items concerning educational equality.All items of the three scales are listed in Table 1 (see also : Schulz et al., 2018).The scale on gender equality explored a mix of attitudes towards women, their choices, and their rights.The scale regarding the rights of racial and ethnic groups focused more strongly on the opportunities that these groups may have in a society, but it was not exclusively focused on the state level.Finally, the scale on the rights for immigrants had some items concerning societal opportunities but focused mostly on rights that the state can bestow ('Immigrants who live in a country for several years should have the opportunity to vote in elections'; see (Schulz et al., 2018).All three scales combine aspects of political and social tolerance.
Due to their conceptualisations, as laid out in the assessment framework, the scales were not optimised to assess students' willingness to engage with marginalised groups.Instead, the scales focused on general opportunities, rights, and responsibilities in society within the dimensions of education, politics, and culture.While the scales covered similar subject matters, few of the items were aligned, so that levels of support for each scale by the different groups cannot be directly compared.To a degree, the variations in the design of the scales reflect the different statuses of the groups.For example, there was no question on educational equality in the scale on the rights of men and women, which may be a reflection of the fact that in many countries, girls score better in academic achievement tests than boys (Rosén et al., 2020).However, the lack of uniformity in the phrasing of the items decreases the comparability of both the items themselves and the scales overall.While the scales of the ICCS 2016 study were equipped to investigate specific aspects of political tolerance, namely tolerance towards social minorities, they lacked consistency in their conceptualisation and phrasing.
For the purpose of further analyses, the two items from each scale that most closely mirrored the definition of political tolerance were selected.These items are shown italicised in Table 1.These items asked most clearly if members of the specific group being

Equal rights for men and women
Men and women should have equal opportunities to take part in government.Men and women should have the same rights in every way.Women should stay out of politics.(-) When there are not many jobs available, men should have more right to a job than women.(-) Men and women should get equal pay when they are doing the same jobs.Men are better qualified to be political leaders than women.(-) Equal rights for ethnic/racial groups All < ethnic/racial groups > should have an equal chance to get a good education in < country of test>.All < ethnic/racial groups > should have an equal chance to get a good job in < country of test>.Schools should teach students to respect < members of all ethnic/racial groups>.< Members of all ethnic/racial groups > should be encouraged to run in elections for political office.< Members of all ethnic/racial groups > should have the same rights and responsibilities.
Equal rights for immigrants < Immigrants > should have the opportunity to continue speaking their own language.< Immigrant > children should have the same opportunity for education that other children in the country have.

< Immigrants > who live in a country for several years should have the opportunity to vote in elections.
< Immigrants > should have the opportunity to continue their own customs and lifestyle.< Immigrants > should have the same rights that everyone else in the country has.
Note. (-) marks inverted items; <> marks phrases that were adapted to the linguistic and cultural context in each country; students were asked how much they agreed or disagreed with the statements; all items could be answered on a 4-point Likert scale from strongly disagree to strongly agree; italicised items were selected for the alignment analyses.
queried should be able to participate in the political process (e.g.'Men and women should have equal opportunities to take part in government') while ignoring items that focused on general attitudes that would fall within the definition of social tolerance (e.g.'Women should stay out of politics').Additionally, only items with a clear connection to political processes were chosen.This way, the scales could reflect students' perspectives on who deserves to participate in political processes.While the items do reflect aspects of political tolerance as defined above, they do so in a narrow way.Decision-making processes can take many different forms that are not covered by the items, such as working within parties, participating in demonstrations or other collective action.Furthermore, the proposed scale only covers three marginalised groups.

Fostering tolerance in schools
As political tolerance is a prerequisite for deliberative processes, it should be of great interest for practitioners and policymakers to know what can be done to impart these values to young people.There is limited research on predictors of political tolerance, especially in adolescents.Janmaat (2022) provides a general overview of research concerning the relationship of education and tolerance and concludes with a call for more academic attention towards political tolerance.Political tolerance is explicitly referred to only in a few studies (e.g., Isac et al., 2019;Maurissen et al., 2018;Sandoval-Hernández et al., 2021).Concepts relating to social tolerance have been applied more often and will therefore be reviewed briefly.Investigating tolerance and intergroup relations among students has a long history in psychological research, and there is good empirical support for the contact hypothesis in particular (Allport, 1971;Pettigrew & Tropp, 2006;Pettigrew et al., 2011).The contact theory suggests that optimal contact between individuals with different group memberships significantly improves attitudes, not only towards the individual with whom the interaction occurs, but also towards the group in general.Optimal contact involves an equal status of participants (such as students), common goals (such as learning or participation), an absence of intergroup competition (e.g. about grades or resources), and should be sanctioned by a person of authority (such as a culturally sensitive teacher).The contact hypothesis has been mostly, but not exclusively, applied to measuring social tolerance.Janmaat (2022), for example, reported that a more diverse classroom composition could foster political tolerance towards different ethnic groups, but the author did not mention studies that use indicators of political tolerance that integrate the support for different groups.Mutz (2002) built on the theoretical and empirical body of the contact hypothesis and explicitly applied it to political tolerance.Mutz proposed that the relationship between exposure to people with dissimilar views and political tolerance is mediated by a cognitive awareness of rationales for different views and by the intimacy of crosscutting associations, i.e. positive connections between people holding those opposing views.According to this theory, people are more likely to develop political tolerance when they are aware of different ideas and the arguments underpinning those ideas, and if they have positive emotional contact with people holding those views.Researchers can now investigate to what extent schools are equipped to foster an understanding of different rationales and relationships between different groups.Political tolerance can be nurtured with cognitive or more social approaches.Cognitive approaches focus on fostering knowledge and an awareness of different perspectives.Schools can create a climate for open discussions in the classroom and thereby promote opportunities for students to learn that multiple, potentially even opposing, perspectives on a societal problem can be valid.Such open discussions have been connected to reduced levels of prejudice at the individual and the classroom level (Miklikowska et al., 2022).Analyses of the Belgian (Flemish) data of the data from ICCS 2009 and 2016 indicated that the perceived fairness of teachers was a significant predictor of tolerance towards migrants (Maurissen et al., 2018).This predictor was significant at the individual and the classroom level.While an open classroom climate for discussion was not predictive of students' tolerance, the values students internalised through participation at school was (Maurissen et al., 2018).On the other hand, three-level analyses of ICCS 2009 data indicated that an open classroom climate was a positive indicator of all three facets of tolerance at the individual and the classroom level (Carrasco & Irrabarra, 2018).Further individual-level analyses of the German ICCS 2016 data robustly suggested that civic knowledge was a strong predictor of tolerance (Ziemes & Abs, 2020).
Developing political tolerance via more emotional and affective means can work by fostering social capital, especially bridging social capital.Social capital can be understood as a resource of individuals and societies alike.Social capital denotes informal relationship structures that can help individuals and groups to achieve goals which they could not otherwise achieve (Coleman, 1988;Putnam, 2000).While bonding social capital connects people within a group, bridging social capital denotes the connection between (members of different) groups (Putnam, 2007).Flanagan (2013) stressed the importance of peer and teacher relationships for the formation of social trust.Flanagan proposed a process of generalisation: if students trust each other, they will also assume they can trust people in general (p.169).In cross-sectional and longitudinal analyses, the perception that teachers were fair and supportive increased trust in people and in political institutions (Abdelzadeh et al., 2015;Lundberg & Abdelzadeh, 2019;Ziemes et al., 2020).Fair treatment by teachers can ensure that all students can voice their opinions and that they are treated properly by students who disagree.
Previous research has shown that tolerance can be influenced by cognitive and social factors.This current research will go beyond previous research by integrating items from different scales on equal rights into one scale of political tolerance.This will make it possible to investigate if a general form of political tolerance can be identified in students in different democracies.This work lays the foundation for exploring which aspects of schooling are relevant to developing political tolerance in a narrower sense.The analyses will distinguish between individual-level and school-level variables and explore if certain variables are relevant only at one level or the other.The analyses will focus on cognitive learning opportunities, an open classroom climate for discussion and teacher-student relationships.As discussed, cognitive learning about a political system might help students understand the relevance of political tolerance within a democracy.The emphasis on cognitive learning is in line with political theories of tolerance and builds on the expectation that an understanding of the system includes an understanding of its prerequisites, such as political tolerance.A climate for open discussions in the classroom can provide students with opportunities to experience stimulating arguments that people from different groups make, and teacher fairness can help foster a productive social climate in which students are more willing to extend tolerance towards different groups.
Classroom discussions and teacher fairness are strongly aligned with psychological theories and criteria of optimal contact (Pettigrew et al., 2011).It is expected that all three aspects of schooling should be positively connected to political tolerance, even after controlling for student background variables.While three-level analyses will be used to identify general trends and to fully take advantage of the sample size, additional twolevel analyses will be employed to ascertain the stability of the three-level analyses and to reveal trends specific to certain educational systems.

Sample
This study uses European data from the ICSS 2016 study.The data is available via the data repository of the International Association for the Evaluation of Educational Achievement (IEA; Köhler et al., 2018).The European data set contains fifteen educational systems with 2,002 schools and 52,788 students.In Belgium, the region of Flanders took part, and in Germany, the federal state of North Rhine-Westphalia (NRW).Both regions have great autonomy over their educational systems.The sample size for each education system is listed in Table 2.The full sample information has been documented in the international and the technical reports respectively (Schulz et al., 2017;Schulz et al., 2018).The school grade targeted included students of about 13.5 years of age on average, which was Grade 8 for most countries.A stratified sampling strategy was used to draw schools and classes within those schools.The goal was to reach representative samples for each educational system.Full classes were assessed; therefore, data analyses need to account for the clustered nature of the sample.In some countries, multiple classes per school were assessed.

Measures
Political tolerance was measured by the items shown italicised in Table 1.Students were asked to indicate their agreement with each items on a 4-point Likert scale.Only the underlined items were used in the statistical analyses.How the alignment method was used to create a scale value is described below.The independent variables were civic learning opportunities, open classroom discussion, student interaction, and teacher-student interaction.
Civic learning opportunities were measured with a scale with seven items that assessed to what extent students had learned something about different civic issues, e.g.how citizens can vote in local or national elections, or how laws are introduced or changed.The scale did not explicitly refer to the importance of political tolerance for deliberative processes but considered general decision-making processes that depend on citizenship participation.It employed a 4-point Likert scale ranging from not at all to to a large extent.
Open classroom climate for discussion is a scale derived from six items concerning teacher and student activities that foster discursive discussions in the classroom.For this context, the most prominent item was 'Teachers encourage students to discuss the issues with people having different opinions' , indicating that the goal was to increase students' understanding of different social and political issues.The items could be answered on a 4-point Likert scale from strongly disagree to strongly agree.
Teacher fairness was assessed with five items measuring teachers' fairness ('Most teachers treat me fairly') and their general relationship with students ('Students get along with most teachers').This scale indicated how good teachers are at building a community with students and at managing students in a just manner.The items could be answered on a 4-point Likert scale from strongly disagree to strongly agree.
The data set contained precalculated item response theory (IRT) scales for the previous three scales.Each scale had a mean around 50 and a standard deviation around 10.The full scales and the results of confirmatory factory analyses of these scales have been documented in the technical report (Schulz et al., 2018).
Gender, immigration history of the family, and socioeconomic status (SES) were used as background variables.The first two variables were especially important as the final scale concerned the rights of migrants and women.Gender was assessed with a single question in which students could indicate if they were a boy (coded as 0) or a girl (coded as 1).Teachers also indicated students' gender, and teacher ratings were used when reports conflicted.Students were assigned an immigration history (coded as 1) when they or both their parents were born outside the country of test.Other students were coded 0. The index for students' SES was comprised of information on books at home, parental education, and parental occupation.SES was part of the data set and calibrated to have a weighted mean of 0 and a standard deviation of 1 within each country.

Procedure
The data was prepared with SPSS 27 and analysed with Mplus 8.7 (IBM, 2021; Muthén & Muthén, 1998-2018).The alignment method was used to construct the scale of political tolerance.The alignment method is a useful procedure when working with many distinct groups and will be shortly summarized based on existing guidelines (Asparouhov & Muthén, 2014;Luong & Flake, 2023;Munck et al., 2017).The alignment method is often seen as either an extension of or as an alternative to a multigroup confirmatory factory analysis.This is a method that tests if a scale measures the same construct in a sufficiently similar manner within different groups (Luong & Flake, 2023).Especially when working with many groups, it is often not realistic to assume full measurement invariance (Munck et al., 2017).Rather than solely examining the similarities and differences in the measurement models in the different groups, the alignment procedure serves to identify a solution with the least amount of non-invariance possible without falling under the fit level of a configural model (Asparouhov & Muthén, 2014).This means that the construct can be assessed by the same items across all groups.Instead of merely assessing levels of non-invariance, the alignment method extracts one or multiple factors in a way that maximises the comparability of the different groups.To achieve this, an algorithm is employed that is comparable to exploratory factory analyses (Asparouhov & Muthén, 2014).This procedure results in factor scores for each case that can be used in subsequent analyses.
This factor score was analysed with multilevel analyses.Multilevel analyses allow the separation of variance between different levels of clusters and account for the clustered nature of data.When participants are clustered in groups such as schools or classrooms, they are more similar than a random sample would be.Not accounting for the clustered nature of data can underestimate standard errors (Gill & Womack, 2013).In ICCS 2016, students were clustered in classes, schools, and countries (Schulz et al., 2018), whereby the school level and the country level were introduced as explicit levels.Furthermore, multilevel analyses allow for the analyses of relationships between different levels.In educational research, it can be of great interest to determine not only to what degree individual perceptions and experiences influence a student's development, but also to what degree the experiences of clusters as a whole are relevant.When variables were used only at the individual level, they were centred around the school group mean.When they were introduced at multiple levels, a latent aggregation was conducted.Both procedures served to purge the student-level analyses of all level-two or above contamination (Marsh et al., 2009).
For multilevel regression analyses including school level predictors, weights were applied in accordance with the guidelines outlined by Rutkowski et al., 2010.At the student level, the student and class weights were combined: WF denotes the general weight factor, WA is the weight adjustment factor that adjusts for non-participation, j represents classes, and i represents students.The school level weights combine the school weight factor with the matching adjustment factor: Analyses solely at the student level employ the student weights provided in the dataset (TOTWGTS).These weights combine the two equations above (see also Schulz et al., 2018).

Multilevel analyses
A baseline analysis was conducted to determine the distribution of variance of political tolerance between the individual, school, and country levels.The analysis revealed that about 6% of the variance was at school level, and about 14% at country level.Therefore, 80% of the variance of political tolerance was located at individual level.Table 2 also reveals that educational systems differ in their school level variance.While for most countries at least 5% of the variance was located at the school level, the same was not true for Croatia.Here only about 2% of the variance was located at the school level.Two sets of analyses will be conducted: First, three level models will be used to identify trends that manifest internationally.Afterwards, multilevel analyses will be conducted for each educational system to investigate the stability of the trends found in the three level approach.Three models were calculated that included the student, school, and the country level.Each analysis adds further complexity.The first model included students' background variables only.The results indicated that girls were more politically tolerant.Other predictors were significant due to the large sample size, but their standardised coefficient was below 0.10 and therefore very low.This model explained about 4% of the variance at the individual level.The second model included learning opportunities and the perceived fairness of teachers on the student level.Girls were still more tolerant in this model.Students' individual learning about politics and the open classroom climate for discussion are significant, but again very small, predictors of tolerance.Teacher fairness was as relevant as students' gender.This model explained about 11% of the variance at the individual level.
The third model is central for the research questions as it introduces variables at the school level.Here, the perceptions of students at a school are pooled and connected to the dependent variable.If students in a school reported experiencing a positive climate for open classroom discussions, they were also much more likely to be politically tolerant.Thus, classroom climate was a significant and strong predictor of political tolerance.Teacher fairness, again as perceived by the students of the school, was also significantly positively connected to political tolerance at school level.The model explained 20.6% of the variance at school level.

Multigroup analyses
As the curricula implemented by different educational systems may vary significantly with a direct impact on students' schooling, additional multilevel analyses were conducted.The third model shown in Table 4 was calculated for each country.Table 2 lists the intraclass correlations for political tolerance by country and already revealed meaningful differences.The intraclass correlations (ICC) ranged from 0.023 in Croatia to 0.123 in Estonia and the Netherlands.The ICC describes how much of the overall variance is located at the school level.The ICC of Croatia would generally be considered small, but even very small ICCs should be considered in analyses (Musca et al., 2011).The school-level variance therefore ranged from rather minor in size to relevant.The results of the two-level analyses are listed in Table 5.In all countries, girls were more tolerant than boys.This effect was strongest in Finland and weakest in Malta.Immigration history of the family and the SES of students were weak predictors of tolerance in most countries.In Belgium, Denmark, Germany (NRW), Lithuania, Norway, and Sweden, the standardised regression coefficients for SES were higher than 0.100, with Sweden at the top with 0.163.In all countries, teacher fairness was the strongest schooling related predictor of tolerance at the individual level, but there was some variance in the relevance of civic learning and an open classroom climate for discussions.At school level, the results resembled the three-level approach.Only in one country (Denmark) was civic learning significantly positively connected to political tolerance.While an open classroom climate was a positive predictor in all countries, it was not significant everywhere.In Lithuania and Malta, the climate for open classroom discussions was an especially strong predictor of tolerance at school level.In most countries, teacher fairness was a positive predictor of tolerance but did not reach statistical significance.
The multigroup analyses indicated that especially the student-level analyses were robust and comparable between countries.The school-level analyses were less stable.This can have multiple reasons: First, the education systems' different approaches to schooling may be impactful here.Some school-level analyses should be interpreted with increased caution.In Croatia, only very little variance of political tolerance was located at school level.Therefore, the strength of the climate for open classroom discussions as a predictor of tolerance was relative as it only explained a tiny portion of the overall variance.

Discussion
This study aimed to shed light on different conceptualisations of social and political tolerance, to discuss the measurement approach taken in ICCS 2016 and to identify aspects of schooling that might be conducive to fostering tolerance.Fostering tolerance can be seen as especially relevant in political systems with prominent populist movements.The literature review suggested the necessity to differentiate between social and political tolerance.While a focus on social tolerance can help to decrease prejudice, political tolerance is more important for political deliberation.Researchers should be aware that further subdimensions exist within these categories, and they should clearly and explicitly position their conceptualisation(s) within their respective study frameworks.ICCS 2016 employed multiple scales that covered aspects of support for equal rights but were not explicitly designed to capture (political) tolerance.The items used did not align neatly with definitions of either social or political tolerance, but the proposed scale can be used to capture a general facet of political tolerance between countries, but not political tolerance in its breadth.
Researchers developing this scale further should be decisive in their theoretical approach and either measure social or political tolerance or, alternatively, acknowledge the multidimensionality of the concept.One way to achieve this might be to incorporate support for the rights of minority groups such as people with disabilities or transgender people, in analogy with concepts such as group-focused enmity (Zick et al., 2008).Using items with more consistent and aligned phrasing for different groups is essential for research that combines items from different aspects of tolerance and may help to make results more comparable.Additionally, items should refer to different aspects of opportunities to participate in society to ensure political tolerance is assessed in its fullest.These procedures would support investigations relating to the question how well students, as emerging citizens, are prepared to take part in deliberative processes with (other) marginalised groups.As it stands, the ICCS framework is also unable to investigate aspects of political tolerance that relate to groups with different political positions.
Intolerance and polarisation between political groups are a challenge for democracies (Reiljan, 2020).
The empirical section of this paper provided multiple insights: First, explicit political learning opportunities are not strongly related to political tolerance.Learning how the democratic political system works does not indicate support for political tolerance even though such tolerance can be seen as a prerequisite for deliberative processes.Second, an open classroom climate for discussion is relevant for political tolerance only if the whole group perceives the climate to be conducive to open discussions.This is in line with previous research (Carrasco & Torres Irribarra, 2018) and hints at the idea that if only specific voices are being heard, discussions will not improve contact or political tolerance.Recent research has also shown that an open classroom climate for discussions at the school level is connected to fewer experiences of vicitimization (Munniksma et al., 2022) and may indeed be part of an democratic school climate that can be fostered by teachers (Maurissen et al., 2018).Third, teacher fairness can be relevant for tolerance at individual and classroom level, indicating the importance of teachers as role moderators and role models.Teacher fairness has reliably been associated with students' political support (Ziemes, 2022).Teacher fairness is part of the school climate and the benefits for the political and academic development are well known (Flanagan, 2013;Thapa et al., 2013), but weakness of the school level is rather surprising.It seems to be the individually perceived respect, not the climate that aids students' formation of political tolerance by fulfilling their individual needs (Mutz, 2002).
The respect students receive from figures of authority is relevant to the rights that students are willing to grant to others.A general climate of fairness also adds to the tolerance afforded to others.When interpreting this last result, one needs to keep in mind that only a fraction of variance was found at school level.Most variance was located at the individual level and was not explained by the variables included in this research.This leads to the fourth conclusion, namely that the analysed educational tools only have a limited ability to foster the political tolerance that democratic systems depend upon.The multigroup analyses showed that schools have some opportunity to foster tolerance at school level, but that additional potential drivers of tolerance need to be identified.Overall, the proposed scale could be a valuable tool for researchers to conduct further secondary analyses concerning political tolerance with ICCS data and beyond.
The results are also relevant for policymakers and practitioners, as they indicate the relevance of individual development trajectories.Fostering political tolerance might be one way to ensure that emerging citizens accept members from different groups as equal participants in the political discourse and to avoid populist polarisation.As the variance was mostly located at the individual level, it is important to ensure that each and every student has the opportunity to take part in discussions and to feel treated fairly by teachers.As the student-teacher relationship was the strongest of the individual-level measures, promoting a positive school climate and good relationships between students and teachers could be one important strategy for fostering tolerance (Rhodes et al., 2009).The statistical irrelevance of civic learning shows that political tolerance is not simply a by-product of general learning about the political system; moreover, it indicates that lessons that explicitly expand on the relevance of political tolerance for the persistence of democracies might be more fruitful (Janmaat, 2022).
This study has multiple limitations.While it was possible to briefly summarise the concepts of social and political tolerance, the format did not allow for an in-depth discussion of tolerance in psychology and political science.Further, the items from ICCS 2016 only assess specific facets of political tolerance, namely the general right of marginalised groups to participate in political processes.Data from fifteen educational systems was used to analyse the role of the school in the development of political tolerance, therefore it was not possible to discuss the context of each educational system, the different systems' approaches to teaching tolerance, or their specific samples.Analyses from the cross-sectional sample can be used to suggest causal associations, but not to prove them.Longitudinal and experimental studies would be necessary to make stronger claims.The present analyses treat the items used as continuous.Yet, some researchers argue that items need at least five categories before they should be regarded as continuous in classical structure equation models (Rhemtulla et al., 2012).As of yet, similar studies do not exist concerning the alignment method, therefore this study employed the procedure laid out by Munck et al. (2017) that treated very similar items as continuous.Further research on the alignment method could add further guidelines concerning the treatment of the prevalent 4-point Likert scales.
The results are mostly in line with findings from previous research, such as those by Carrasco and Irrabarra (2018), Maurissen et al. (2018), and Ziemes and Abs (2020).Yet, the present study goes beyond previous research by creating a unified scale of political tolerance and by using the alignment method to ensure its international comparability and by combining three-level with multigroup analyses.While the three-level analyses made full use of the sample size, especially at the school level, the two-level analyses were apt to identify differences between education systems.Finally, the present study adds to previous research by comparing multiple predictors of tolerance in different education systems while regarding the student and the school level.Overall, the study illustrates the value of conceptual clarity regarding political and social tolerance for research.Additionally, the alignment procedure has proven to be a valuable tool for finding statistical solutions that sufficiently approach invariance for international analyses.Finally, the results of the investigation underscore the importance of teacher fairness and classroom discussions as perceived by all students in a school for developing political tolerance.

Table 2
Descriptive statistics of the data set (Schulz et al., 2018)ical report(Schulz et al., 2018); 2 means weighted at country level (TOTWGTS); 3 anchor country in the fixed alignment analysis; 4 Germany (NRW) reached one third of the intended sample.
Table 3 presents the results of the alignment analyses.The bold indicators in brackets indicate non-invariant parameters.

Table 3
Invariance analysis of political tolerance Countries with significantly non-invariant measurement parameters are shown within parentheses; country codes are listed in Table2. Note.

Table 4
Predictors of political tolerance at individual and school level Note.Estimator = MLR; Standardised coefficients; ***p < .001

Table 5
Weighted multigroup multilevel analysis of political tolerance regressed on aspects of individual background and schooling