Abstract
Civic attitudes are crucial for a well-functioning democracy. Research shows these attitudes are influenced by adolescents’ families and peers. However, little is known about the interplay between these socializing agents. This study examines the role of similarity between parents’, friends’, and classmates’ trust in groups, institutions, and sources of information and positive attitudes toward people with a migration background. A total of 540 eighth-grade students (Mage = 14.0) from 32 Dutch schools participated. Parents’ and classmates’ civic attitudes, but not friends’ attitudes, were positively related to adolescents’ attitudes regardless of the perceived quality of the relationship. The results support the congruency hypothesis, that more similarity in civic attitudes of meaningful others strengthens socialization effects. These insights deepen our understanding of adolescents’ civic attitudes.
Introduction
Adolescence is an important period for the development of civic attitudes (e.g., Prior, 2010; Russo & Stattin, 2017). Recently, studies have indicated a weakening connection between youth and democratic systems. For example, the most recent International Civic and Citizenship Education Study [ICCS, 2022] showed that although adolescents still largely embrace democratic values, anti-immigrant attitudes are a topic of concern. The percentage of adolescents that disagree with the statement that people with a migration background should have the same rights that everyone else in the country has stagnated between 2016 and 2022 (Schulz et al., 2023), while the population has become more multicultural in most countries.
Empirical research on adolescents’ attitudes has stressed the since long (cf. Bandura, 1977; Bronfenbrenner, 1979; Coleman, 1988; Merton, 1957) assumed importance of attitudes of meaningful others in the formation of adolescents’ attitudes. Of all the different settings which adolescents are part of, the home, peer, and school contexts are considered particularly important socializing contexts. Indeed, adolescents’ civic attitudes have consistently been found to be related to the attitudes of their parents and peers (e.g., Degner & Dalege, 2013; Miklikowska et al., 2019; Wanders et al., 2021; Zagrean et al., 2022). Although there is scholarly consensus that adolescents develop civic attitudes within the social contexts in which they are embedded, empirical research about the interrelatedness of the different contexts is less common (Bobba et al., 2024).
For some adolescents, the home, peer, and school contexts are congruent: there is high similarity in prevailing attitudes. In this study, we argue that including degree of similarity in civic attitudes of parents, friends, and classmates leads to a more profound understanding of these attitudes held by adolescents. We focus on two civic attitudes among Dutch adolescents: positive attitudes toward people with a migration background and trust in institutions. By considering the congruence between socializing agents, we aim to contribute to a deeper understanding of how various contexts contribute to young citizens’ formation of these civic attitudes.
Bronfenbrenner’s Ecological System Theory
Bronfenbrenner’s (1979) ecological system theory assumes that human development happens in interaction with the individual’s social environment. This environment consists of different contexts that are related to the individual and to each other. Bronfenbrenner has developed and changed the theory in later phases (Rosa & Tudge, 2013), but the main principle was always the ecological paradigm: human development is viewed as emerging from the interactions between individuals and the contexts in which they are situated (Merçon-Vargas et al., 2020; Navarro et al., 2022). A central mechanism in the theory of Bronfenbrenner is the principle of synergy: “the cooperative action of discrete agencies such that the total effect is greater than the sum of two or more effects taken independently” (Bronfenbrenner & Morris, 2006, p. 800).
Taking Bronfenbrenner’s ecological system theory as a starting point for understanding the role of contextual factors in the development of adolescents’ civic attitudes, it is assumed that the different contexts in which adolescents are embedded, interact with each other in relation to adolescents’ civic attitudes. Previous studies have given important insights into the independent relations of different socializing agents and adolescents’ attitudes, showing the relative importance of each aspect of the social network (Degner & Dalege, 2013; Dostie-Goulet, 2009; Koskimaa & Rapeli, 2015; Rossi et al., 2016; Wanders et al., 2021). This field of research consistently shows the socializing role of school, parents, and peers in adolescents’ civic development. Experiences within these different social contexts all contribute to shaping adolescents’ views on society.
Parents
Parents are often considered to be principal agents of socialization in childhood (Degner & Dalege, 2013). They contribute to the development of their children’s values in various meaningful ways, both directly and indirectly (for an overview, see the studies by Grusec & Davidov, 2019; Spinrad et al., 2019). Parental socialization and the internalization of parents’ values have been described in various theories, most explicitly in the social learning theory (Bandura, 1977) and parenting styles (Baumrind, 1971). From the perspective of the social learning theory (Bandura, 1977), parents serve as role models for their children, and children learn through observating and experiencing their behavior (e.g., Bricheno & Thornton, 2007; Le et al., 2024). Role models are significant others whose attitudes or behaviors act as an example. They provide a model upon which someone bases their ideas about determining appropriate thoughts or actions (Merton, 1957). As role models, parents share their own ideologies and values through their behavior, practices, and communication (Kelly, 2006). Thereby, parents provide norms, manage children’s environments, and reinforce specific attitudes (Miklikowska et al., 2019). To a certain degree, adolescents may have internalized these attitudes, values, and beliefs, resulting in parent-child similarity (Degner & Dalege, 2013). In this regard, adolescents’ interpretation or understanding of these norms, perceived norms (Lapinski & Rimal, 2005; Rimal & Lapinski, 2015), is also of interest, since how adolescents perceive their parents’ behavior and beliefs plays an important role in the socialization process (Hatemi & Ojeda, 2021; Stattin et al., 2022).
In this way, parents play an important role in the development of adolescents’ civic attitudes and behavior, both explicitly and implicitly. They influence youths’ civic behaviors with their own world views as reflected in their behavior, practices, and communication (Fletcher et al., 2000; Kelly, 2006; White, 2021). In a meta-analysis of parent-child similarity in intergroup attitudes, Degner and Dalege (2013) find that parent-child attitudes are related throughout childhood and adolescence. Parent-child similarity has also been found for political alienation (Gniewosz et al., 2009). Altogether, it is expected that adolescents’ civic attitudes are related to their parents’ civic attitudes.
Peers
During adolescence, peers become increasingly important in the development of adolescents’ attitudes and behavior (for overviews, see the studies by Brown & Larson, 2009; Giletta et al., 2021; Laursen & Veenstra, 2021), whereas the relative influence of parents typically decreases (De Goede et al., 2009). Early and mid-adolescence is seen as a period in which autonomy and individuation in the parent-child relationship increases, and identity formation plays a major role. This phase coincides with rapid neurological, cognitive, and social changes. The process of self-determination and the exploration of one’s own identity and of the surrounding world and its dominant norms, as well as the uncertainties associated with this processes, make adolescents receptive to support and guidance from peers. In this process, comparison with and confirmation by relevant others through behavior, opinions, and norms in the group of peers play an important role through observation and coding of meanings. Processes such as the pursuit of homophily, confirmation, and reward, for example, inclusion, admiration, and more generally, the acquisition of status and prevention of rejection and loss of status, influence self-definition, self-affirmation, and social acceptance and provide clues for socially desirable outcomes, leading to acceptance among peers as influential significant others. Also with regard to civic development, the importance of peers increases during adolescence. A recent, large-scale longitudinal study into the development of democratic values among Dutch adolescents shows that in the final years of secondary school, between the ages of 16 and 18, adolescents indicate learning most about politics and society from their friends—compared to parents and teachers—and also talk about politics and social issues with them more often than before (Huijsmans et al., 2024). In the present study, we distinguish between friends within the classroom and all classmates to examine different peer influences.
Among peers, friends are considered to be especially important socializing agents. Friends are choosen voluntary, and friendships can be ended one-sided, making affirmation and attention to the dynamics of the relationship important features. This leads to investing in the relationship through, for instance, openness toward ideas, resulting in friendship as a powerful channel for development and identity formation. Friends are usually the ones with whom adolescents spend much time with, which provides many opportunities for observational learning and social interaction. Following social learning theory, friends act as role models who influence adolescents’ attitudes through information exchange, modeling, and reinforcement. For example, adolescents observe the behavior of their friends, which may introduce them to new viewpoints or the consequences of such opinions (Ryan et al., 2019). Individuals are more likely to be a resource for observational learning when they are more similar and share characteristics. Moreover, adolescents may be more inclined to internalize attitudes of friends from the desire to maintain the social relationships (Laursen & Veenstra, 2021; Zingora et al., 2020).
Empirical research supports the role of adolescents’ friends in the development of civic attitudes. In a study on social interest among adolescents, Wanders and colleagues (2021) find that awareness of social issues raised by friends becomes increasingly important during adolescence. Moreover, friends can increase one’s xenophobia or, on the other hand, promote tolerance toward people with a migration background (Van Zalk et al., 2013).
Finally, among peers, classmates form an important social reference group for adolescents (Thijs & Verkuyten, 2013). With regard to citizenship competences, the classroom is often considered to be a crucial context (e.g., Rossi et al., 2016). The classroom, or school, can be seen as a mini polity (Flanagan et al., 1998) because in this social setting, students can develop a sense of what it means to be a member of a community, and they can put citizenship into practice. Moreover, classmates can also be role models, as being surrounded by peers with certain attitudes stimulates the adoption of these attitudes through the process of social learning (Busching & Krahé, 2020). Research into peer influence on civic attitudes shows positive relations between adolescents’ attitudes and those of their peers in the same school, for example, with regard to attitudes toward people with migration background (Miklikowska et al., 2019) and prosocial behavior (Busching & Krahé, 2020).
Relationship Quality
An important factor in the mechanisms of these socialization processes is the (perceived) quality of the relationship (Grusec et al., 2000). Developmental and psychological theories in the line of attachment research (Bowlby, 1951) suggest that secure attachment motivates children to agree with their parents and internalize their attitudes. Recent empirical research underlines the effect of the relationship quality. For example, the study of Miklikowska (2016) shows that the parent-adolescence correspondence was higher when adolescents perceived their parents as being more supportive. Adolescents need to feel that their parents are supportive and understanding to comply with parental socialization efforts (for an overview, see the studies by Grusec, 2011; Smetana et al., 2015).
Analogous to parents, perceived support from peers may be related to the strength of the socialization effect: more perceived support from friends may be linked to more openness toward friends’ ideas, which stimulates the formation of similar attitudes. Within the classroom, a sense of community can strengthen socialization effects. For example, research shows that the link between adolescents’ multicultural beliefs and the beliefs of their classmates is stronger when adolescents feel more accepted by their classmates (Thijs & Verkuyten, 2013).
Contextual Congruence
In previous research on different socializing agents, the focus is often on the relative importance of each context (Amnå et al., 2009), thereby focusing on the relative impact of each agent on adolescents. However, according to Bronfenbrenner (1979), the role of environmental factors in development is best understood when looking at the synergy between the various contexts: these factors should be understood in interaction with each other. The contextual-congruence model (Spera & Matto, 2007), which is also based on the ecological system theory, states that a high level of congruence in a child’s socialization network facilitates positive child development outcomes and leads to greater internalization of socialization messages. In this theory, degree of congruence is defined as “the level of match or similarity between the values, goals, expectations, and aspirations advocated for and acted upon by stakeholders across the child’s different and unique mesosystems” (Spera & Matto, 2007, p. 554). The assumption that follows from this is that more congruence in socializing agents’ attitudes strengthens the relation between those attitudes and adolescents’ civic attitudes.
The idea that congruency in social contexts has implications for the formation of one’s orientations and values is also central in research into the effects of social capital in education. In 1987, Coleman and Hoffer postulated the functional community hypothesis, which boils down to the idea that more congruent social contexts have a positive effect on students’ academic outcomes. According to this hypothesis, a more connected network (i.e., more social relationships between parents and teachers, and between parents, classmates, and peers) forms a functional community (Coleman & Hoffer, 1987) with positive effects on access to information, norm reinforcement, and effective value transmission. Morgan and Sørensen (1999) add that such a collective mechanism is mainly to be expected for reinforcement of values (norm enforcing social capital), but not so much for new information (horizon expanding social capital). Coleman and Hoffer formulate a similar assumption with regard to value communities, where, in addition to the effect of close social ties, the idea of value congruence is central. Shared values between important others lead to a value community around the adolescent, with positive effects on the transfer of dominant values. Studying secondary schools, Dijkstra et al. (2004) report confirmation of such community effects for schools with students from religious communities in the Netherlands. Also for students’ citizenship competences, such an effect of religious schools with a distinct religious (i.e., orthodox Protestant) profile was found (Dijkstra et al., 2023).
With regard to civic attitudes, some studies highlight the importance of examining interrelated effects of different socializing agents. In a study into the effects of parents, peers, and classroom diversity on the development of adolescents’ prejudice, an interrelated effect was found showing that parental attitudes predicted the attitudes of peers (Miklikowska et al., 2019). Bobba and colleagues (2024) found that parents’ and classmates’ unique influences interact in predicting adolescents’ prejudice. Furthermore, in a study on youth civic engagement, Zaff and colleagues (2008) use a systematic approach by proposing the idea of a “civic context,” including parent and peer influences, ethnic and cultural traditions, and religious practices. Their study shows that peer, parent, and cultural factors cumulatively result in higher levels of civic activities into adulthood.
Understanding the effect of congruence in socializing contexts is important during adolescence. Not only is adolescence seen as a critical period for the development of civic attitudes (e.g., Prior, 2010; Rekker et al., 2015; Russo & Stattin, 2017), it is also a period that is characterized by changes in parent-child and peer relationships. The importance of peers in the socialization process increases, and the parent-child relationship typically changes from more dependent and controlling to increased autonomy (Giletta et al., 2021; Lionetti et al., 2019). Moreover, it is a period in which individuals (further) develop cognitive abilities, such as reasoning and metacognitive skills. This enables them to acknowledge and contemplate multiple points of view, replacing mono-dimensional perspectives (Daniel et al., 2012; Rest et al., 2000). Altogether, especially during adolescence, congruence in attitudes of adolescents’ meaningful others may become more important (Spera & Matto, 2007). Investigating congruence in relation to adolescents’ civic attitudes will thereby increase our understanding of the role of meaningful others in this developmental period.
Present Study
The present study investigates congruency of three interrelated social contexts of adolescents: civic attitudes of parents, friends, and classmates and their relation to civic attitudes of adolescents. The following research question is studied: To what extent is the relation between civic attitudes of adolescents and the civic attitudes of their parents, friends, and classmates strengthened by the degree of congruency between these attitudes and by the quality of each relationship?
We will test the following hypotheses: (a) Civic attitudes of adolescents are positively related to civic attitudes of their parents, friends, and classmates; (b) the relation between civic attitudes of adolescents and the attitudes of their parents, peers, and classmates is moderated by the quality of each relationship; and (c) civic attitudes of adolescents are more strongly related to civic attitudes of their parents, friends, and classmates when the attitudes of parents, friends, and classmates are more similar.
The study focuses on the Netherlands. Similar to most European countries, schools in the Netherlands are statutory obligated to foster citizenship as part of the curriculum and school culture by promoting students’ support for fundamental democratic values and developing students’ citizenship competences. However, promoting citizenship education in the Netherlands has been challenging, and results did not meet the anticipated outcomes (De Groot et al., 2022; Inspectorate of Education, 2016). The ICCS 2022 results for the Netherlands showed that Dutch adolescents in general had less positive attitudes toward people with a migration background, in comparison to the international average, as was the case in 2016. Trust in civic institutions was higher among Dutch adolescents than the international average but lower than that in more similar countries in terms of the human development index and democracy (Daas et al., 2024).
Methods
Registration
The study protocol was preregistered with Open Science Framework (OSF) prior to data analyzing on March 20, 2023 (https://osf.io/7p8qa).
Participants and Procedure
A subset of the Dutch dataset from the International Civic and Citizenship Education Study (ICCS) 2022 was used (Schulz et al., 2023). ICCS is a cross-sectional, international study on lower-secondary-school citizenship education and students’ citizenship knowledge, attitudes, and behaviors. Target population of ICCS were adolescents enrolled in the eighth grade, which is the second grade of secondary school in the Netherlands (13–14 years old). Schools were randomly selected by a two-staged stratified sampling design (Schulz et al., 2023). When a school agreed to participate, one class was randomly selected. All students within this class were asked to participate. Adolescents were excluded from participating in the study if their parents or they themselves objected to participate.
Data were collected between March and July 2022. All questionnaires were filled out in the classroom setting. Participants did not receive any compensation, and schools received a report with the (anonymized) average results of their school. More information about the procedure of the ICCS data collection can be found in the international (Schulz et al., 2023) and the Dutch national reports (Daas et al., 2024).
For the current study, schools participating in the randomized ICCS sample in the Netherlands were invited to take part in an additional, online questionnaire. The same class of students filled in the online questionnaire, which included questions about the perceived relationship quality and civic attitudes of parents and peers. After data cleaning, 540 students of 32 schools (of the 124 Dutch schools in the ICCS sample) were included in the dataset. The mean age of the participants was 14.0 years (SD = 0.5). Of the participants, 9.6% had a migration background, and 52.2% of the participants identified as boy, 46.5% as girl, and 1.3% as other. The 32 schools are a fairly representative sample of the Dutch school population in terms of key characteristics. For example, the distribution of the educational tracks in the subsample is comparable to the complete Dutch ICCS sample, which was stratified to equally represent the different educational tracks.
Before data cleaning, 667 students of 33 schools were included in the data. There was one class of which all students were excluded because none of the students answered the question about friends within the classroom. Specifically, during data cleaning, participants were excluded from the analyses if (a) data were unidentifiable due to technical errors (11 participants), (b) adolescents only participated in the additional questionnaire, not in ICCS (52 participants), (c) adolescents did not answer the question about friends within the classroom (55 participants), (d) unreasonable duration, more than 30 minutes or less than 2 minutes (five participants), and (e) invalid answers to the question about which parent filled out the questionnaire (four participants), for example, “my dog.”
Variables
Data from ICCS and the additional questionnaire were used. When deviating from the initial ICCS scale structure (e.g., a different selection of items), factor analysis was conducted to examine the underlying structure of the items. Otherwise, the mean of the items or the initial scale score was used (reported alphas are based on the sample used in this study).
Dependent Variables
Adolescents’ positive attitudes toward people with a migration background was measured by ICCS by five items asking adolescents to what extent they agree or disagree with statements about equal rights and cultural, social, and economic benefits of immigrants, ranging from 1 (Strongly agree) to 4 (Strongly disagree). An example item is “Immigrant children should have the same opportunities for education that other children in the country have.” The mean of the five items (Cronbach’s alpha = .85) was used in the analyses.
Adolescents’ trust in groups, institutions, and sources of information was measured in ICCS by asking participants to what extent they trust certain groups, institutions, and sources of information, ranging from 1 (Completely) to 4 (Not at all). Fifteen items were included (Cronbach’s alpha = .88). An example item of groups is “Scientists,” an example item of institutions is “Courts of justice,” and an example item of sources is “Traditional media (television, newspaper, radio).” The items were combined by using confirmatory factor analyses (CFAs; see Statistical Analyses).
Independent Variables
Civic attitudes of classmates were assessed in ICCS, since all students within the same class were asked to participate. The class averages of positive attitudes toward people with a migration background and trust in groups, institutions, and sources of information were used in the analyses. To minimize data reduction, all scores of classmates participating in ICCS were used to calculate the mean score, including the scores of students who did not participate in the additional questionnaire.
Civic attitudes of friends in the classroom were also assessed in ICCS. Participants were asked in the additional questionnaire to fill out one to three names of classmates they spend most time with since the start of the school year. These names were used to match the questionnaires with the ICCS data. Names were replaced by anonymous identification numbers and removed from both datasets.
Civic attitudes of parents were filled in by adolescents, by asking them to fill out to what extent they think their parents agree with certain statements. These questions were part of the additional questionnaire. Before answering these questions, adolescents were asked which parent or caregiver they spend most time with. Answer options included father, mother, stepfather, stepmother, foster father, foster mother, grandfather, grandmother, or someone else. The questions about parents’ attitudes were then answered about the selected parent or caregiver. To optimize comparability with the answers of the adolescents, items identical to ICCS were used (while adjusting the subject, i.e., “you” into “your”). For the questions about positive attitudes toward people with a migration background, all five ICCS items were used (Cronbach’s alpha = .88). For the questions about trust toward groups, institutions, and sources, next to a selection of six ICCS items, two additional, general items were used: trust in society and trust in democracy. Two important considerations in deciding to deviate from the ICCS scale were the length of the original scale and the specific questions that were considered difficult to answer for someone else (i.e., their parent or caregiver). CFA was used to combine the eight items into one scale (Cronbach’s alpha = .77).
Quality of the relationship with parents was measured in the additional questionnaire by the responsiveness subscale (eight items) of the Nijmeegse Rearing Questionnaire (NRQ, Gerris et al., 1993). Participants were asked to what extent they agree with statements about themselves and their parent on a 6-point scale ranging from 1 (Totally disagree) to 6 (Totally agree). An example item is: “My knows very well how I feel.” The mean of these scores (Cronbach’s alpha = .98) was used in the analyses.
Quality of the relationship with friends was measured in the additional questionnaire by combining five items from the scale Support of the Network of Relationship Inventory (NRI, Furman & Buhrmester, 1985) and two adjusted items of the NRQ. An example is: “How much do you care about these friends?” Answer options of the NRI ranged from 1 (Not at all) to 5 (A lot). To create a scale indicating the quality of the relationship among friends, the items of the NRI and NRQ were combined (Cronbach’s alpha = .89). Given the different length of Likert-type scales (5-point and 6-point scales), scores were standardized to a mean of 0 and standard deviation of 1, before calculating the mean. The mean of the standardized scores was used in further analyses.
Adolescents’ perceptions of student interaction at school were assessed in ICCS and used as an indicator of quality of the relationship with classmates. This scale consists of three items (Cronbach’s alpha = .77). An example item is: “Most students at my school get along well with each other.” The ICCS scale score is used in the analyses.
Control Variables
Adolescents’ gender, migration background, and socioeconomic status were included as control variables, as there are indications that these affect the variables of interest (e.g., Daniel et al., 2012; Wang & Benner, 2016). All control variables were assessed in ICCS. Gender was assessed by asking students “What is your gender?,” with answer options “boy,” “girl,” or “other.” Migration background was a dichotomous variable, assessed by asking adolescents in which country they and each of their parents were born. When both parents were not born in the Netherlands, the answer was recoded to “immigrant background.” Finally, socioeconomic status was based on parental occupation, parental education, and number of books in the home (Schulz et al., 2023).
Statistical Analyses
Before conducting the analyses, items were recoded so that higher scores reflect more trust, a more positive attitude toward people with a migration background, and higher quality of the relationship. CFAs 1 in Mplus were conducted to test whether the items of the dependent variables loaded on a single factor. The one-factor model including the five items measuring positive attitudes toward people with a migration background showed good model fit, both for adolescents (Root Mean Square Error of Approximation [RMSEA] = .08, Comparative Fit Index [CFI] = .98, Standardized Root Mean Squared Residual [SRMR] = .02) and for parents (RMSEA = .07, CFI = .99, SRMR = 0.02). The mean score of the items was used in the analyses. The initial one-factor model for trust in institutions, groups, and sources did not fit well. Based on modification indices, covariances between items were added step-by-step to ensure good model fit. Seven covariances were added, and this model showed satisfactory fit (RMSEA = .08, CFI = .92, SRMR = .05). For parents, only one adjustment (covariance between media and social media) was needed (RMSEA = .07, CFI = .95, SRMR = .04). Factor scores were exported and used in further analyses.
Intraclass correlation coefficients (ICCs) were calculated to check for the variation in civic attitudes between schools. The ICC for positive attitudes toward people with a migration background was .07, indicating that 7% of the total variance of positive attitudes toward people with a migration background is between schools. Although this is a relatively small percentage, the model with the random intercept fits significantly better than a model without a random intercept (L-ratio = 8.92, p = .003). For trust in groups, institutions, and scources, the ICC was also .07. Again, the model with the random intercept fits better than a model without a random intercept (L-ratio = 7.66, p = .006). In conclusion, multilevel analyses are conducted to account for the nested structure of the data.
To test Hypotheses A and B, multilevel regression analyses were conducted. Adolescents’ civic attitudes were included as dependent variables. To test whether civic attitudes of adolescents are positively related to civic attitudes of consecutively their parents, friends, and classmates, civic attitudes of parents, friends, and classmates were used as independent variables. To test Hypothesis B, interactions of the perceived quality of the relation and civic attitudes for all three socializing agents (parents, friends, and classmates) were included to examine the possible moderating effects. Multilevel regression analyses were conducted in R version 4.3.2, using the multilevel package (Bliese, 2022). Participants with missing values on the dependent, independent, or control variables and one school with a very small number of students (N < 5) were excluded from the analyses. All independent variables were grand mean centered to ease interpretation, except the average of friends’ attitudes. This variable was group mean centered, in order to distinguish within- and between-class effects (Feaster et al., 2011). Assumptions for the multilevel analysis were checked. Variance Inflation Factor [VIF] values indicated no multicollinearity. Besides some outliers, residuals were approximately normally distributed. On the second level, signs of non-normally distributed residual errors were found. However, this is relatively common and appears to have little or no effect on the parameter estimates (Maas & Hox, 2004).
To test Hypothesis C, discrepancies were calculated. First, the variance between the attitudes of parents and friends and the class average were calculated for each adolescent. Second, the difference between the adolescent’s attitudes and the attitudes of parents, friends, and classmates were calculated. This resulted in three discrepancy scores. In the final step, the correlations between the variance and the three difference scores were estimated.
Results
Descriptives
The means, standard deviations, and correlations between all variables are shown in Table 1. Most adolescents (68.3%) reported to spend most time with their mother, and therefore filled out the questions about her.
Descriptive Statistics and Correlations.
p < .05. **p < .01.
Multilevel Results
To test Hypotheses A and B, multilevel analyses were conducted. One class was excluded because of a very small number of participating students (N = 3). After exclusion, 31 schools were included in the analyses. Hypotheses were tested by running seven consecutive models with stepwise introduction of groups of variables. Models were run separately for positive attitudes toward people with a migration background (Table 2, N = 506) and for trust in groups, institutions, and sources (Table 3, N = 503).
Positive Attitudes Toward People With a Migration Background Regressed on Individual and Classroom Variables.
Note. u0: variance on classroom level; R: variance on individual-level, immigration background: 0 = other, 1 = yes.
p < .05. **p < .001.
Trust in Groups, Institutions, and Sources Regressed on Individual and Classroom Variables.
Note. Values are given as regression coefficients and standard deviations. u0: variance on classroom level; R: variance on individual-level, immigration background: 0 = other, 1 = yes.
p < .05, **p < .001.
Positive Attitudes Toward People With a Migration Background
In the first model, friends’ attitudes and parents’ attitudes were added as individual-level variables (Level 1). In this model, the perceived attitudes of parents is a significant predictor of adolescents’ attitudes: When parents have a more positive attitude toward people with a migration background, adolescents also have a more positive attitude. The average attitudes of friends is not a significant predictor.
In the second model, individual background variables (gender, migration background, socioeconomic status) were added as control variables, which improves the model fit significantly. In the third model, the classroom average of positive attitudes toward people with a migration background was added as classroom-level variable (Level 2), which is a significant positive predictor of adolescents’ attitudes. Adding the classroom average results in a significant improvement of the model fit.
In the fourth model, the indicators of quality of the relationship with parents and friends were added. Both are non-significant predictors, although the model fit improves significantly. In the fifth model, the interaction effects on the individual level were added to examine the moderation effect of quality of the relationship on the effect of friends’ attitudes. This does not significantly improve the model, and the interaction variables are non-significant.
In the sixth model, the indicator of quality of relationships with classmates was added, showing a small, significant effect and significant improvement of the model fit. Finally, in the seventh model, an interaction effect on the classroom level was added to examine the moderation effect of the perception of student interactions in the school on the effect of classroom average attitudes. Adding the interaction effect does not improve the model fit in comparison to the previous model.
In summary, the multilevel analyses show that parents’ positive attitudes toward people with a migration background and the classroom average are positively related to adolescents’ positive attitudes toward people with a migration background, while controlling for gender, socioeconomic status, and migration background. These relations are not moderated by the perceived qualities of relationships. Attitudes of friends within the classroom are not significantly related to adolescents’ attitudes.
Trust in Groups, Institutions, and Sources of Information
In the first model, friends’ attitudes and parents’ attitudes were added as individual-level variables (Level 1). As shown in Table 3, in this model, parents’ trust was a significant predictor of adolescents’ trust: When parents have more trust in groups, institutions, and sources of information, so do adolescents. The average of friends’ trust was not a significant predictor.
In the second model, individual background variables (gender, immigration background, socioeconomic status) were added as control variables, which improves the model fit significantly. Third, the classroom average of trust was added as classroom-level variable (Level 2), which is a significant predictor of adolescents’ trust. The model fit improves in comparison to the previous model.
In the fourth model, the indicators of quality of the relationship with parents and friends were added. Both indicators are non-significant, although the model fit improves. When adding the interaction variables (Model 5), the model fit improves. The interaction for parents’ trust is significant, indicating that the effects of parents’ trust is moderated by the quality of the relationship as experienced by the adolescent. The interaction for friends’ trust is not significant.
In the sixth model, the indicator of quality of relationships with classmates was added, which is a significant predictor. Again, adding this variable improves model fit. However, the interaction of parents’ trust and perceived relationship quality is no longer significant, whereas the interaction of friends’ average trust and perceived relationship quality is now a significant predictor. Finally, in the seventh model, an interaction variable at the classroom level was added, to examine the moderation effect of perception of student interactions in the school. The effect is non-significant, and the model fit does not significantly improve.
To sum up, the multilevel analyses show that perceived parent’s trust in groups, institutions, and sources and the classroom average of trust in groups, institutions, and sources are positively related to adolescents’ own trust in groups, institutions, and sources of information, while controlling for adolescents’ gender, socioeconomic status, and immigration background. Perceived quality of the relationship with friends moderates the association with friends’ average trust, although the effect is small, and the main effects are non-significant.
Discrepancy Scores
To examine whether greater similarity in attitudes of parents, friends, and classmates is associated with greater similarity between the attitudes of adolescents and their respective parent, friends, and classmates (Hypothesis C), correlations between the variance between attitudes of parents, friends, and the classroom average and the three discrepancy scores (indicating the difference between adolescents’ attitudes and the attitudes of each socializing agent) were estimated. The results, as displayed in Table 4, show weak to moderate positive correlations. This indicates that when attitudes of adolescents’ parent, friends, and classmates are more similar, the attitudes of the adolescents are also more similar to those of their parent, friends, and classmates. This finding is consistent for all two-way combinations of parent, friends, and classmates and for both attitudes toward people with a migration background and for trust in groups, institutions, and sources of information.
Correlations for Positive Attitudes Toward People With a Migration Background in the Upper Right and for Trust in Groups, Institutions, and Sources of Information in the Lower Left.
Class average includes scores of friends and adolescents. Without these scores, the correlation is still significant but less strong.
Significant at p < .01.
Discussion
Multiple developmental theories discussed in this article suggest that adolescents develop civic attitudes through interaction with the social contexts in which they are embedded. For example, they discuss social issues with friends, practice how to resolve conflicts at home, and learn about democracy at school. In this study, we expand upon studies showing the distinct or relative effects of different socializing contexts on adolescents’ civic attitudes, by examining whether the relation between adolescents’ civic attitudes and the civic attitudes of their parents, friends, and classmates is strengthened by the degree of congruence in attitudes and by the quality of each relationship. By investigating these joint effects, we aim to improve understanding of the relation between adolescents’ civic attitudes and their social context.
The results of the multilevel analyses partially supported Hypothesis A, both for positive attitudes toward people with a migration background and for trust in groups, institutions, and sources of information. Civic attitudes of adolescents were positively related to civic attitudes of their parents and classmates, but not to the civic attitudes of their friends within the classroom, while controlling for socioeconomic status, gender, and migration background.
We found almost no evidence supporting Hypothesis B. We observed only one small interaction effect between two factors that do not have significant direct effects: average of friends’ trust in groups, institutions, and sources of information and perceived quality of the relationship. We did not find the same moderating effects for friends’ positive attitudes toward people with a migration background. The relation between civic attitudes of adolescents and the attitudes of their parents and classmates is not moderated by the quality of these relationships. More specifically, we found that the moderating effect of parent’s relationship quality on the association between parents’ trust and adolescents’ trust disappeared when including interaction with classmates. This suggests that the attitude of classmates and the interaction with them are stronger factors in explaining adolescents’ trust, which may overshadow the importance of the relationship quality of parents and their values. Moreover, the results show that for adolescents’ parents and classmates, these civic attitudes are related to their attitudes regardless the quality of the relationship. Although the relative importance of classmates with regard to parents is in line with previous research, as the influence of parents typically decreases during adolescence and peers become more influential for one’s attitudes (De Goede et al., 2009; Huijsmans et al., 2024), the finding that relationship quality did not have a moderating effect for parents and peers was unexpected. In other studies, a postive moderating effect was found (Miklikowska, 2016; Thijs & Verkuyten, 2013). However, it is possible that our findings indicate that adolescents already experienced more similarities and a relatively positive quality of the relationship with the parent or friends about whom they answered the questions, as they were asked to answer about the persons they spent most time with. According to social learning theory, these are important mechanisms in observational learning. Future research could shed more light on the aspect of similarity in the socialization process of civic attitudes by including both the duration and amount of time spent with parents and peers.
The results highlight the importance of parents and classmates in adolescents’ civic socialization. At this age, parents’ attitudes as perceived by adolescents are important for their own civic attitudes. This outcome is consistent with previous research (e.g., Jennings et al., 2009; Miklikowska, 2016; Wanders et al., 2021). The strong association with the average of classmates’ attitudes that was found in our study is also in line with prior research that has highlighted the importance of peer socialization in adolescence (e.g., Giletta et al., 2021; Thijs & Verkuyten, 2013; Zingora et al., 2020). In terms of the social contexts in which adolescents are embedded, both home and school appear to be important for adolescents’ civic attitudes: the attitudes that parents and classmates hold are apparently meaningful for the way adolescents develop their outlook to the world.
However, this does not apply to friends within the classroom: the average of their civic attitudes was not significantly related to adolescents’ attitudes. There are several possible explanations for this unexpected finding. First, we had information about the classmates with whom the adolescents spend most of their time. There may be other, stronger mechanisms in peer socialization within the class that affect adolescents’ internalization of peer attitudes. For example, an important aspect may be the social group norms, which are most strongly influenced by the most popular classmates (Laursen & Veenstra, 2021; Zingora et al., 2020). Second, because of the design of our study with one sampled class per school, only friends within the classroom were included in our study. Assuming that friendship is characterized, among other things, by voluntariness and intimacy, friendships might also come from other, pre-existing social contexts. It is therefore plausible that the restriction of friendships within the classroom leads to selection, leaving out friends with whom the closest relationships are maintained, so that the effect of friendship cannot be sufficiently estimated. But even then, the role of friends in civic socialization may still be limited in early adolescence. A longitudinal study among Dutch adolescents showed that in the second year of secondary school, young people discuss political and social issues mainly with parents, compared to friends and teachers (Van der Meer et al., 2021). From age 16 onwards, the role of friends becomes more important (Huijsmans et al., 2024; see also the study by Wanders et al., 2021).
Finally, our study shows that civic attitudes of adolescents are more strongly related to civic attitudes of their parents, friends, and classmates when the attitudes of parents, friends, and classmates are more similar (Hypothesis C). This finding supports the idea that congruence in social contexts strengthens socialization processes, as postulated in Bronfenbrenner’s (1979) ecological system theory, the functional community hypothesis (Coleman & Hoffer, 1987), and the contextual-congruence theory (Spera & Matto, 2007). Thus, not only the relative importance of the different social contexts matter, there is also a joint effect of the various contexts.
Suggestions for Further Research
While we have argued that congruence in civic attitudes of meaningful others strengthens the socialization process, this does not necessarily imply that such congruence is preferable for adolescents’ civic development. For example, adolescents may also benefit from interactions with different values, as it could stimulate them to reflect upon their own attitudes and prepare them to engage in a diverse society. In line with Bronfenbrenner’s ecological system theory, our study specifically focused on understanding the mechanism of congruence in the socialization process, rather than on the (normative) implications. Whether the effect of congruence is desirable may also depend what these attitudes entail. This could be an important direction for future research.
Moreover, as our study implies that congruence matters in socialization processes, this also poses the question how incongruence in social contexts relates to adolescents’ civic attitudes. Previous studies show that adolescents’ response to incongruence is not determined or unambiguous but can vary. Looking into engagement with school and learning, Phelan and colleagues (1991) propose the Multiple Worlds model which focuses on interrelationships between the values, beliefs, and expectations of students’ family, peers, and school. For some students experiencing incongruent settings, the differences were seen as too big to overcome, which could result in frequent conflict or disengagement from one setting. Other students were able to manage these boundary crossings. Also regarding socialization of values, there has been research on how adolescents cope with disagreements among socializing agents’ values. Daniel and colleagues (2012) argue that besides preferring one set of values over the others or integrating influences of different contexts into one coherent set, adolescents may also accept incompatible values as they are, holding divergent value priorities relevant to different life contexts (Daniel et al., 2012). Examining the effect of incongruence in social contexts on the development of civic attitudes during adolescence is an important direction for future (longitudinal) research. In addition, although the focus is often on influence of socializing agents, the socialization process is assumed to be bidirectional, as adolescents themselves have an active role in it (Grusec, 2011). In summary, from the perspective that attitude formation is inherently social and dialogical, the question of how adolescents make sense of the conflicting civic attitudes they encounter deserves more specific attention.
Limitations
There are some limitations of our study that should be addressed. First, in the measurement of friends’ attitudes, congruence between attitudes may also be caused by selection effects, as adolescents tend to form friendships with peers who have similar attitudes (Zingora et al., 2020). Furthermore, congruence between attitudes of classmates may be caused by shared experiences rather than peer socialization. This links to a second limitation of the study, the cross-sectional data. We were not able to investigate causality or socialization processes during adolescence. As Ryan and colleagues (2019) stated, longitudinal data are needed to disentangle selection from socialization. Moreover, longitudinal data would enable to investigate bidirectionality: Adolescents may also actively influence their parents and peers in their civic attitudes. Since we used cross-sectional data, we were unable to examine the full range of interactions as proposed in the ecological system theory of Bronfenbrenner, such as the effect of time.
Implications for Practice
Notwithstanding the aforementioned limitations, this study has shown that the socialization effects of different social contexts should not be considered in isolation. On the contrary, congruence matters: Similarity in civic attitudes of parents, friends, and classmates strengthens the assocations between their attitudes and those of adolescents. Understanding this collective mechanism of socialization processes contributes to theories on the development of citizenship competences. To understand why adolescents hold certain civic attitudes, it is essential to consider the social contexts in which they are situated.
Moreover, our contextualist orientation may help to better connect with the experiences of young people. It implies that, when promoting adolescents’ civic attitudes, one should be considerate of the fact that these attitudes are affected by the different social contexts in which they are embedded. Making sense of different civic attitudes of meaningful others can be challenging for adolescents. However, being aware and respectful of values different than your own is an important aspect of being a democratic citizen. Schools in particular can have a unique role in supporting adolescents to interpret and respond to different civic attitudes, especially as schools potentially reach all adolescents. In this social setting, adolescents often encounter attitudes that differ from those in their home context. One implication for educational practice could be to provide opportunities to have conversations about these differences. In its function as the “secondary apprenticeship system” (Ten Dam et al., 2004), schools may provide a context to reflect on (in)congruence together and offer possibilities for personal meaning-making. For teachers, being aware of social norms and social dynamics within students’ own contexts is essential in this. Making use of the pedagogical space of the school, teachers can help students to cross boundaries and connect citizenship practices at school with socializing contexts outside school. In conclusion, by acknowledging the importance of attitudes of meaningful others and the possible differences in contexts, adolescents’ civic development can be better understood to support the broadening of their horizons.
Footnotes
Disposition editor: Cristina Mogro-Wilson
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by the Nationaal Regieorgaan Onderwijsonderzoek (NRO, 40.5.19340.001).
