1 Introduction

Civic engagement has recently been the focus of a new and intense research debate given its importance for developing socially and politically active citizens (Adler & Goggin, 2005). Some research has identified a decline in civic engagement –especially among adolescents– through indicators such as the intention to vote or concerns about social issues (Dudley & Gitelson, 2002; Oosterhoff et al., 2020; Putnam, 2000).

Despite a lack of consensus about the reported decline, concern increases regarding adolescent citizen involvement given that youths’ altruistic ideologies towards social problems generally lead to active forms of social participation in adulthood (Putnam, 2000), and active and involved citizens are fundamental for maintaining a democratic society. Therefore, it is crucial to understand which factors promote civic engagement during this developmental stage. The present study aims to explore how perceived support from the adolescents’ environmental contexts contributes to developing civic engagement –assessed through the expectation of future sociopolitical participation– by fostering a sense of belonging to a larger, dependable, and stable structure: sense of unity.

1.1 Expected Sociopolitical Participation as an Indicator of Civic Engagement

Civic engagement has been approached from different perspectives, such as the political, educational, or informal (Sherrod et al., 2010). This diversity broadens the definition however makes it difficult to arrive at a consensus.

Karakos (2015) proposes three fundamental aspects for defining civic engagement: a context (social, developmental, political, etc.); a community relationship; and its occurrence for pro-social reasons in benefit of the community. Along these lines, the International Civic and Citizenship Education Study (ICCS) defines adolescent civic engagement as the attitudes and behaviors –along with the expectations of future participation– related to a general and manifest social participation (Schulz et al., 2010). These authors note that adolescent civic engagement includes indicators of: (1) psychological participation, understood as the youths’ beliefs and thoughts surrounding their intervention in society; (2) individual civic participation, through debates or seeking out information on political and social issues; (3) youth participation in activities –both in and out of school– related to collective civic engagement; and (4) the adolescent’s intention of future political participation. This present study focuses on the latter which, given adolescents’ limited opportunities for participation, is considered to be an important indicator of civic engagement. The expectation of sociopolitical participation refers to the behavioral intentions of sociopolitical participation, including forms of social engagement (e.g. volunteering), political participation (e.g. interest in collaborating with a political party), unconventional forms of participation (e.g. participating in protests), and collaborating with special interest groups (e.g. animal protection associations) (Flanagan et al., 2007).

1.2 Sense of Unity as a Precursor of Sociopolitical Participation

To understand which factors can foster adolescent sociopolitical participation it is key to considerer that adolescent development takes place within social groups and communities in which they learn to be aware of and value the common good. Community connection contributes to developing relationships that are beneficial for the youth’s adaptive development (Lerner et al., 2014), and furthermore, it is this sense of community which leads to common interests and actions that foster the adolescent’s sociopolitical participation (Schulz et al., 2010). Sense of community has been defined as “the perception of similarity to others, an acknowledged interdependence with others, a willingness to maintain this interdependence by giving to or doing for others what one expects from them, the feeling that one is part of a larger dependable and stable structure” (Sarason, 1974, p:157). However, previous studies vary substantially on the operationalization of this construct and its conceptualization, using terms such as sense of community or connectedness to measure different qualities of social relationships (including feelings, attitudes, behaviors, etc.) and contexts in which these social relationships take place –from more specific groups of belonging (such as the family or the school) to a broader understanding of social connectedness to “others” (Barber and Schluterman, 2008).

The present research focuses on sense of unity –a new concept closely related to the aforementioned–, understood as a perceived similarity to others which creates recognized interdependence, a willingness to perpetuate it, and the feeling of belonging to a larger, reliable, and stable structure. Our study uses the scale proposed by the SALUD project (Stimulating Adolescent Life Skills Through Unity and Drive; Samdal et al., 2016) for the Health Behaviour in School-aged Children (HBSC) study. This scale is based on an instrument proposed by Larson (2006), employed to evaluate the positive feeling derived from belonging to a large social structure. Sense of unity offers common benefits, positive adolescent development, and adds the approach of universal values for successful socialization of youth into responsible and competent citizens, related to understanding, appreciation, tolerance, and protection for the welfare of all people and not just their closest community members.

1.3 The Influence of the Adolescent’s Developmental Contexts on Sense of Unity and Expected Sociopolitical Participation

Sense of community, belonging, or sense of unity –as employed in the present work–, are predictors of sociopolitical participation and are strongly determined by the individual’s developmental contexts, offering opportunities to improve prosocial behaviors towards their community (Albanesi et al., 2007; Encina & Berger, 2021). By adopting an ecological perspective this phenomenon can be understood in a developmental and cultural framework (Ben-Arieh & Attar-Schwartz, 2013). The present research focuses on the microsystemic social-structure as defined by Bronfenbrenner (1979) –represented by the family, friends, and school–, and examines how support from these developmental contexts relates to their sociopolitical participation through the overall sense of unity.

The family is the primary socializing context, in which strong bonds are established based on caring and trust (Lamb & Lewis, 2011). In addition, family is a source of social capital –understood as norms, institutions, and organizations that foster trust and cooperation among people, communities, and society– as well as playing a fundamental role in accessing other forms of social capital through social networks, the neighborhood, or school (Collins & Laursen, 2004; Steinberg & Silk, 2002). Existing research therefore emphasizes the family’s role in youth civic engagement by instilling caregiving values in the family members (Kim et al., 2015). However, Alesina and Giuliano (2011) identified an inverse relationship between civic and political engagement and trust in the family as a provider of services and resources. According to these authors, strong family ties tend to substitute rather than complement social trust, producing a type of amoral familiarity in which trust and caregiving are found only within the family.

Secondly, activities promoted or implemented within the school microsystem could foster a democratic climate and have an important influence on both the adolescents’ present and future civic engagement (Kim et al., 2015). Schools and teachers offer youth a conducive context for expressing opinions, debating, and participating in activities related to school government (voting for delegates, student council, etc.), all of which predict future commitments towards their community (Flanagan et al., 2007). In this regard, McFarland and Thomas (2006) found that adolescent participation in activities both in and out of school predicted voting as adults. The school context is also a privileged place for developing a sense of belonging, by promoting high-quality interpersonal relationships, connectedness, respect for diversity, and community partnerships (Wang & Degol, 2016). Moreover, teacher characteristics such as encouraging student participation in class, enthusiasm for teaching, class organization, and kindness foster a higher perception of teacher support among students, thereby increasing the adolescent’s sense of belonging (Chiu and Churchill, 2016; Freeman et al., 2007).

Lastly, peer influence is especially relevant on the microsystem level –especially during adolescence– due to social developments in addition to physical and cognitive changes. Being accepted by friends and schoolmates is essential for positive adolescent development (Laursen, 2017). Likewise, Wray-Lake and Shubert (2019) found peer political discussion to have a positive influence on adolescent civic engagement. Similarly, friends support and communication have been related to both present and future adolescent political participation (Zaff et al., 2003). In addition, adolescents who discuss politics with their peers have shown higher expectations of participating in community service activities, whereas spending unproductive time with friends has been related to lower civic knowledge and less support of minority rights (Wilkenfeld, 2009).

1.4 The Role of Sex, Age, and Socioeconomic Status in the Contribution of Different Developmental Contexts to Adolescents’ Expected Sociopolitical Participation through the Sense of Unity.

Sex, age, and socioeconomic status are relevant sources of inequalities in adolescent sociopolitical participation. Previous research have found sex differences in adolescent sociopolitical participation, for example whereas girls are more likely to engage in social issues and organizations, boys are more likely to engage in political activities (Jenkins, 2005). Along these lines, Cicognani et al. (2012) found boys to be more interested in politics and girls to be more socially engaged, however no sex differences were identified in civic or political participation.

In addition, considering the developmental perspective, previous findings have shown differences throughout the different stages of adolescence. Researchers have found civic engagement to increase with age, as does identity formation, autonomy, and opportunities for participation (Flanagan, 2004; Hardy et al., 2014). Moral identity and autonomy in decision making –factors strongly associated with social responsibility (Wray-Lake & Syvertsen, 2011)– increase with age and therefore enhance adolescents’ prosocial behaviors (Eisenberg et al., 2006).

Lastly, family socioeconomic status is another source of inequalities in adolescent sociopolitical participation. Families with a better socioeconomic position are more likely to equip adolescents with a broader range of social skills and educational resources higher in political content, thus promoting a greater diversity of civic experiences (Flanagan et al., 2009; Zaff et al., 2009). In addition, families with a high-socioeconomic level are better able to cope with the costs derived from sociopolitical participation (Lenzi et al., 2012).

Furthermore, previous findings –although scarce and inconsistent– have found sex, age, and socioeconomic status to affect the sense of community. For example, whereas Albanesi et al. (2007) found no sex differences, other research shows that adolescent boys have a stronger sense of community than girls (Chiessi et al., 2010; Cicognani et al., 2012). Similarly, Chiessi et al. (2010) found no age differences in adolescents’ sense of community, whereas Albanesi et al. (2007) found that it decreases as adolescents age. Regarding socioeconomic status, Vieno et al. (2005) found socioeconomic differences in adolescents’ sense of community, however it was only significant regarding school level and not individually. Finally, previous studies have also reported sex, age, and socioeconomic differences in perceived support from families, friends, teachers, and classmates (Cheng & Chan, 2004; Due et al., 2003; Gecková et al., 2003).

Despite the well-documented importance of developmental contexts during teen years, their influence in political socialization has been studied and compared infrequently. As an example, Dostie-Goulet (2009) found that during this stage peer influence becomes increasingly more relevant for adolescents’ political interests while parental influence declines. The present research adds to the existing evidence an analysis of the different influences of family, friends, teachers, and classmates on adolescents' expected sociopolitical participation. In addition, although some research has been conducted analyzing the sense of community or belonging to a specific context in adolescents’ political participation, this study examined the role of the sense of unity as a broader concept mediating this association. Moreover, given evidences of sex, age, and socioeconomic inequalities in perceived support, sense of community or belonging, and sociopolitical participation, this study examined if there is a similar contribution of perceived support from the different contexts to sociopolitical participation through the sense of unity for boys and girls, adolescents with different ages, and pertaining to different socioeconomic positions. Therefore, the aims of the present study are: (1) to analyze the influence of perceived support from developmental contexts (family, friends, teacher, and classmates) on adolescents’ expectations of future sociopolitical participation through the sense of unity, and (2) to evaluate the invariance of the model according to sex, age, and socioeconomic status.

2 Method

2.1 Study Design and Participants

This research was conducted in the framework of the OPINA Barometer (Barómetro OPINA) project (more information in www.barometro-opina.es), carried out by a research team from the University of Seville (Spain) in collaboration with UNICEF. The transversal study evaluated the adolescents’ opinions and concerns, their knowledge about sociopolitical issues, their implication as citizens, as well as their wellbeing.

The collected data comes from a representative sample of more than 8,500 adolescents between 11 and 18 years old and guaranty an estimation error lower than 1% (99% estimation accuracy) with a 95% confidence interval. Participants were selected using multistage random sampling stratified by conglomerates according to age, region, habitat (urban or rural), and the type of school (public or private). Given that some questions are sensitive to age, three versions of the questionnaire were employed to match the age-groups (11–12 years old, 13–14 years old, and adolescents 15 years old or older). For the present study, 3,715 participants 13 years old or older were selected (53.1% girls). Regarding age groups, 51.4% were 13–14 years old, 39.0% were 15–16 years old, and 9.6% were 17–18 years old.

Data was collected through an anonymous online questionnaire administered at school. The use of Information and Communication Technologies (ICT) avoids the use of paper, facilitates the student’s participation, and has more guaranties for anonymity and confidentiality, as well as avoids human errors in data entry.

Therefore, school collaboration consisted in facilitate student-participant access to computer labs where, under teacher supervision, they have completed the questionnaire. Three basic conditions were met throughout data collection: (1) the students have responded to the questions by themselves; (2) the responses are anonymous; and (3) the questionnaire was administered at school.

Schools were first contacted by phone and later sent relevant information via email. Once confirming their participation, they were sent instructions for accessing the questionnaire, consent form, system requirements, and recommendations for the teachers who would be supervising the students. Once data collection was finished, each school was sent an individualized infograph of the principal results.

Responses to the questions about sex, age, and academic year were required, so all participants presented valid responses. However, the rest of the questions allowed participants to omit responses. A minimum 50% of questionnaire completion was required in order to examine response quality for each subject according to missing values and omitted questions. An algorithm was created to examine response reliability by including analysis of coherence in different questions or an abusive use of extreme values. After applying the aforementioned criteria for guaranteeing participant response validation, the final sample of study participants was created and later that of this article.

The questionnaire was approved by the University of Seville’s Ethical Committee (Comité Ético de Experimentación de la Universidad de Sevilla) in accordance with the standards of the 1,964 Helsinki Declaration and its later modifications. Informed consent was obtained from the school staff, parents/legal guardians, and students.

2.2 Instruments

The instrument used was the Opinion Barometer of Childhood and Adolescence (Barómetro de Opinión de la Infancia y la Adolescencia) (Moreno et al., 2017). In addition to sex and age, the following variables were selected from the questionnaire:

  • Family socioeconomic level was evaluated through the latest version of the 6-item instrument Family Affluence Scale (FAS-III) (Torsheim et al., 2016). The items assess family material affluence through ownership of certain goods such as the number of cars, computers, or bathrooms. The Cronbach alpha coefficient of this scale was 0.48. These results are congruent with previous psychometric validation of different versions of FAS that revealed weak to moderate loadings onto a common factor, however, showing high test–retest reliability (r = 0.90) and consistency between child and parent reports (r = 0.80) (Torsheim et al., 2016). In this research, FAS-III was employed as a categorical variable, distributing the subjects into three groups: the highest 20% classified as high-affluence; the lowest 20% as low-affluence; and the middle 60% as medium-affluence, as recommended in the last HBSC report (Inchley et al., 2020).

  • Civic engagement was assessed through the scale Expectations of sociopolitical participation in the future (ESPP). The scale is a new proposal used in the OPINA Barometer, composed of seven items combining the three items of the Expectations for Engagement in Community Issues scale, one item from the Expectations for Engagement in Electoral Politics scale, one item from the Expectations for Unconventional Political Engagement scale and two items adapted from the Endorsement of Special Interest Groups scale, all used previously and validated, showing alpha values ranging between 0.72 and 0.80 (Flanagan et al., 2007). The items evaluated the likelihood of adolescent participation in each of the actions in the future through a Likert scale with values between 1 (not at all likely) to 5 (extremely likely). The alpha coefficient in this study of the full scale was 0,76. Analysis of the unidimensional latent structure showed an adequate fit of the data (CFI: 0.95; RMSEA = 0.08; SRMR = 0.05).

  • Sense of unity (SU) was evaluated through a new measure developed by SALUD project (Stimulating Adolescent Life Skills Through Unity and Drive) and included in the Health Behaviour in School-aged Children (HBSC) study (Samdal et al., 2017). The scale is composed of eight items, and is based on prior research on sense of community lead by Sarason (1974), Barber & Schluterman (2008), and Larson (2006). Some of the items included in this scale are: “I feel a strong sense of togetherness”, “I feel responsibility for others”, and “I feel that it is good to be part of a community”. This new scale has been piloted in quantitative and qualitative studies among Norwegian students (Samdal et al., 2017). In the quantitative study, based on adolescents aged 15–16 (n = 2,240), the scale showed a Cronbach's alpha of 0.89 and meaningful correlations with indicators of mental health, social support, and school. The qualitative study –in the form of a classroom discussion about all the items (based on 28 Norwegian students aged 13–14 years)– indicated high validity and a good understanding of all questions. The Cronbach alpha coefficient for this study was 0.76, showing adequate internal consistency. Goodness of fit index indicated an adequate fit of the data to a unidimensional latent structure (CFI: 0.99; RMSEA = 0.04; SRMR = 0.03).

    In addition, variables related to the adolescents’ perception of support in their different developmental contexts were included. The following instruments were used to measure the quality of the relationships in each context:

  • Family support (FS) was evaluated through the family subscale of the Multidimensional Scale of Perceived Social Support (MSPSS; Zimet, et al., 1988). The subscale is composed of the following four items: “My family really tries to help me”, “I get the emotional help and support I need from my family”, “I can talk about my problems with my family”, and “My family is willing to help me make decisions”. The responses ranged on a 7-point Likert scale from 1 (very strongly disagree) to 7 (very strongly agree). The alpha confidence in the original study was 0.87. In this study, the Cronbach alpha was 0.91 and fit indices for the model showed good internal structure (CFI: 0.99; RMSEA = 0.03; SRMR = 0.02).

  • Friends support (FRS) was evaluated through the friends subscale of the Multidimensional Scale of Perceived Social Support (MSPSS; Zimet et al., 1988). This subscale consists of four items: “My friends really try to help me”, “I can count on my friends when things go wrong”, “I have friends with whom I can share my joys and sorrows”, and “I can talk about my problems with my friends”. Responses ranged on a 7-point Likert scale from 1 (very strongly disagree) to 7 (very strongly agree). The alpha confidence in the original study was 0.85 indicating adequate internal consistency (Zimet et al., 1988). In the present study, this was confirmed with a Cronbach alpha of 0.91 and fit indices for the model showing an excellent internal unidimensional structure of the subscale (CFI: 0.99; RMSEA = 0.04; SRMR = 0.02). Several previous studies have confirmed high internal consistency, reliability and discriminant validity of both subscales of family and friends support (Dahlem et al., 1991; Edwards, 2004).

  • Teacher support (TS) and classmates support (CS) were evaluated using the two respective subscales from the Perceived Support from Teachers and Classmates Scale, which was developed and validated within the international HBSC network, showing the two-factors model structure of the scale (Torsheim et al., 2000). Teacher support was measured by three items: “I feel that my teachers accept me as I am”, “I feel that my teachers care about me as a person”, and “I feel a lot of trust in my teachers”. The classmate support consisted in a scale composed of 3 items: “The students in my class enjoy being together”, “Most of the students in my class are kind and helpful”, and “Other students accept me as I am”. The responses of both scales were recorded on a 5-point Likert scale from 1 (strongly disagree) to 5 (strongly agree). In the original study, both scales were validated in a cross-national sample of 13- and 15-year-old adolescents, including an additional item in each one and demonstrating good internal reliability. Alpha values ranged from 0.77 to 0.81 for teacher support, and from 0.74–0.75 for classmate support (Torsheim et al., 2000). Further development and validation have been carried out in different countries within the HBSC network (Freeman et al., 2017), which support having reduced the scale to three items as was used in this research. The scale shows internal reliability with Cronbach values of 0.82 and 0.74 for the teacher and classmates’ subscales, respectively. The Cronbach alpha values for this study were 0.82 and 0.92 for each scale, respectively. Goodness of fit indices indicated an adequate model fit (CFI: 0.97; NNFI: 0.97; RMSEA = 0.08; 90% CI: 0.07,0.09; SRMR = 0.03).

2.3 Data Analysis

Firstly, descriptive statistics were used to examine sample distribution according to the studied variables –including the mean, standard deviation, and minimum and maximum values–, and Pearson’s coefficient correlation to examine the associations between them.

Secondly, the Student t-test and ANOVA with a Bonferroni post hoc test were employed to examine mean differences in all variables across sex, age, and socioeconomic groups, also estimating the effect size of the differences with Cohen’s d. Effect size was interpreted following the established criteria in behavioral sciences: small effect for values around 0.30, moderate effect for values between 0.30 to 0.50, and strong effect when values were equal to or higher than 0.50 (Cohen, 1988). IBM SPSS Statistics 26.0 was used to analyze descriptive statistics, correlations, and mean comparisons with a minimum confidence level of 95%.

Thirdly, a structural equation model was used to analyze how the adolescent’s perception of support from each developmental context (family, friends, classmates, and teachers) influences their expectations of sociopolitical participation in the future through their sense of unity.

Lastly, the estimation method of maximum likelihood (ML) and the Chi-squared test (χ2) were used to test the general model adjustment. A model provides an adequate fit of the data when Chi-squared is not significant. However, given that large sample sizes increase its value, other indexes were also included: Comparative Fit Index (CFI), Root Mean Square Error of Approximation (RMSEA), and Standardized Root Mean Squared Residual (SRMR). For the CFI, acceptable values are considered as those above 0.90 and excellent those above 0.95. For RMSEA and SRMR indices, values close to 0.08 and lower than 0.05 were proposed as indicators of good model adjustment (Hu and Bentler, 1999). Moreover, the bootstrapping procedure (1,000 bootstrap samples) was performed to examine the significance of the total direct and indirect effects by analyzing the unstandardized coefficients (95% confidence intervals) and the standardized coefficients to evaluate the strength of the association and its effect size. To test this effect size, and following Peterson and Brown (2005), the standardized β coefficients were transformed into r and this indicator in eta-square (η2) following Dunlap’s procedure (1994). This statistic was interpreted as indicating a small effect for values around 0.05, moderate effect for values 0.06 to 0.11, and as strong effect when values were equal to or higher than 0.14 (Cohen, 1988).

Finally, the model’s measurement invariance across the variables sex, age, and socioeconomic status was analyzed. Configural invariance without any parameter restriction was tested, and path diagrams with the standardized coefficients of the most relevant direct and indirect model is presented segmented by sex, age, and socioeconomic status. In addition, the Z Fisher test was employed to examine significant variations in the indicators from the different groups. Only significant differences are reported. Lastly, path coefficients were constrained to test metric invariance. Given the Chi-squared statistic’s sensitivity to sample size, an increase in CFI higher than 0.01 was considered to be an indicator of a significant change in the model by sex, age, and FAS (Cheung & Rensvold, 2002). The statistical software program JASP 0.14.1, based on R, was used for examining the latent factors through confirmatory factorial analysis (CFA), testing the mediation model and the significance of the indirect effects through the bootstrapping procedure, and to analyze the measurement invariance.

3 Results

3.1 Descriptive Statistics and Correlations

Table 1 presents the descriptive statistics, including minimum, maximum, mean, and standard deviation values, for continuous variables.

Table 1 Descriptive characteristics of the sample population in all the analyzed variables (n = 3,715 adolescents 13–18 years old)
Full size table

Pearson’s correlation coefficient values (Table 2) showed that all indicators of developmental contexts (family, friends, teacher, and classmates support) have positive and significant relationships (p < 0.05) with the scores in ESPP and SU, showing associations with a low to moderate intensity (except for the association between CS and ESPP, which was not significant).

Table 2 Pearson’s correlation coefficients between indicators of developmental contexts, expected sociopolitical participation, and sense of unity
Full size table

The associations between perceived support from the different contexts and ESPP were lower (the r value oscillating between 0.03 and 0.08) than their association with SU (the r value oscillating between 0.22 and 0.32). In addition, associations amongst the different contexts were also low to moderate (with the r value oscillating between 0.23 and 0.37), demonstrating no collinearity between them.

3.2 Sex, Age, and Socioeconomic Differences in Perceived Support from the Different Contexts, Expected Sociopolitical Participation, and Sense of Unity

As shown in Table 3, results from the Student’s t test showed significant sex differences for CS and FRS, as well as for ESPP and SU. Boys showed higher scores in CS than girls, whereas girls showed higher FRS compared to boys. However, in both cases, the effect size of the differences was small. Girls also showed higher scores than boys in ESPP and SU, with these differences being moderate and small, respectively.

Table 3 Mean comparisons between girls and boys and measurement of the effect size
Full size table

Regarding age differences (Table 4), comparison among the three groups (13–14, 15–16, and 17–18 years-old) using the ANOVA test only showed significant differences in FS (F (3713, 2) = 3.23, p = 0.39), however with a negligible effect size (d = 0.09), and with no significant differences between age groups in the Bonferroni post-hoc comparisons.

Table 4 Comparison analysis of means by age groups of the different variables
Full size table

With respect to socioeconomic status, the ANOVA test used to compare the three groups according to their family affluence showed significant differences in all variables, except for TS (F (3713, 2) = 2.27, p = 0.103; d = 0.06) and SU (F (3713, 2) = 0.119, p = 0.888; d = 0).

Therefore, adolescents pertaining to families with high FAS reported higher FS (F (3713, 2) = 34.16, p = 0.001), CS (F (3713, 2) = 18.41, p = 0.001), and FRS (F (3713, 2) = 12.74, p = 0.001) than adolescents with medium and low family affluence. In addition, adolescents with medium family affluence also showed higher support in the three variables than those with low family affluence. Results from the Bonferroni post hoc test showing the significance for the comparison of the three groups and all effect sizes are presented in Table 5.

Table 5 Comparison analysis of means by family socioeconomic level of the different variables
Full size table

Lastly, differences among the socioeconomic groups were also found for ESPP (F (3713,2) = 2.26, p = 0.103). However, the effect size test and the post hoc comparisons revealed that the differences among the three groups were negligible (d values around 0.3 and 0.12).

3.3 Testing a Mediation Model of the Influence of Perceived Support from the Developmental Contexts on Expected Sociopolitical Participation through the Sense of Unity and its Invariance across Sex, Age, and Socioeconomic Status

As presented in Table 6, the goodness of fit indices of the hypothesized mediation model showed good fit to the data (CFI = 0.923; RMSEA = 0.056; SRMR = 0.046) despite a significant Chi-square (χ2 = 3942.83; p < 0.001).

Table 6 Goodness of fit indices for the proposed factorial model and the invariance analysis
Full size table

Figure 1 presents the standardized coefficients of the baseline model estimating the direct association between perceived support from the different contexts and SU, and between SU and ESSP in the global sample, as well as the indirect effects of support from the different contexts on ESPP through SU.

Fig. 1
figure 1

Representation of the standardized estimations of the path coefficients of the global model

Full size image

The model explained 11% of the variance of ESPP and 22% of the variance of SU. The direct paths from support from the different contexts to ESPP showed low factor loadings and were not significant. However, mediation effects assessed using the bootstrapping procedure suggested that the effect of perceived support from the developmental contexts on ESPP operates through SU. Specifically, FRS (B = 0.06; 95% CI = 0.04, 0.07; p < 0.001; β = 0.09; η2 = 0.02) and CS (B = 0.07; 95% CI = 0.05, 0.10; p < 0.001; β = 0.07; η2 = 0.01) showed stronger indirect effects (with a small effect size) on ESPP through SU than FS (B = 0.03; 95% CI = 0.01, 0.04; p < 0.001; β = 0.05; η2 = 0.01) and TS (B = 0.55; 95% CI = 0.04, 0.07; p < 0.001; β = 0.03; η2 = 0.01).

Regarding the direct effects from the support from the different contexts on SU, results showed FRS (B = 0.15; 95% CI = 0.12, 0.18; p < 0.001; β = 0.26; η2 = 0.10) and CS (B = 0.21; 95% CI = 0.15, 0.28; p < 0.001; β = 0.20; η2 = 0.08) to have moderate effect sizes and stronger effects than FS (B = 0.03; 95% CI = 0.00, 0.06; p = 0.007; β = 0.07; η2 = 0.01) and TS (B = 0.09; 95% CI = 0.04, 0.13; p < 0.001; β = 0.09; η2 = 0.02). The direct effect from SU to ESPP was also significant with a strong effect (B = 0.34; 95% CI = 0.28, 0.40; p < 0.001; β = 0.35; η2 = 0.16).

To test the model’s measurement invariance, configural invariance –without parameter constrains– was first performed segmenting the sample by sex, age, and socioeconomic status (see Figs. 2, 3, and 4 respectively). This was then repeated testing metric invariance through parameter constrains across the different groups (as shown in Table 6).

Fig. 2
figure 2

Standardized solution of the final model across the different groups by sex. The data is presented as boys/girls

Full size image
Fig. 3
figure 3

Standardized solution of the final model across the different groups by age. The data is presented as 13-14/15-16/17-18 age groups

Full size image
Fig. 4
figure 4

Standardized solution of the final model across the different groups by socioeconomic level. The data is presented as low/medium/high socioeconomic level

Full size image

Regarding results of configurational invariance, Fig. 2 presents the standardized coefficients for both models of boys and girls. The models explained 11% (for boys) and 8.9% (for girls) of ESPP, and 23% (for boys) and 20.6% (for girls) of SU. The small variations among the standardized coefficients were examined using Fisher’s Z test, showing that changes were not significant (p < 0.05). Therefore, the model revealed configurational invariance across sex.

Figure 3 presents the model without parameter restrictions across age. The model explained 12.1% (in 13–14-year-old adolescents), 11% (for 15–16-year-olds), and 8.3% (in 17–18-year-olds) of ESPP. In addition, the model explained 19.8% of SU for 13–14-year-old adolescents, 22.7% of 15–16-year-olds, and 35.2% for 17–18-year-olds.

The direct association between perceived support from the different contexts and ESPP showed low and insignificant loadings, similar to the general model. However, the indirect effects on ESPP through SU were significant in all the age groups with a small effect size in the majority. Examining in more detail the differences among the parameters across age using the Fisher’s Z test showed a significant change only in the indirect effect from FS to ESPP through SU, being stronger in the 17–18 year-old group (β = 0.06, p < 0.001; η2 = 0.01) in comparison with the 13–14 (β = 0.01, p < 0.001; η2 = 0.003; Z = -2.27, p = 0.011) and 15–16 (β = 0.02, p < 0.001; η2 = 0.005; Z = -1.884, p = 0.03) year-old groups.

Regarding analysis of the direct effects of the different contexts on SU and the direct effect from SU on ESPP, all parameters were similar to the global model, with no significant differences by age, as supported by the Fisher’s Z test (p < 0.005). Therefore, the model revealed configurational invariance across age.

In Fig. 4, the direct and indirect effects are examined in the three groups according to socioeconomic status, without parameter constraints. The model explained 12.2%, 10%, and 11.5% of ESPP, and 32.7%, 23%, and 17% of SU in adolescents with low, medium, and high family affluence, respectively. Fisher’s Z test showed only three significant variations.

Firstly, the direct effect of FS on SU in the low family-affluence group was significantly different when compared to the medium family-affluence group (Z = 2.79, p = 0.003). Specifically, this effect showed a moderate effect size in the low FAS group (β = 0.13, p = 0.015; η2 = 0.03), being negligible in the medium FAS group (β = 0.00, p = 0.886; η2 = 0). For adolescents pertaining to families with high FAS (β = 0.09, p = 0.039; η2 = 0.01) the effect size indicated a low intensity.

Secondly, the direct effect of CS on SU showed a significant change, being stronger in the low FAS group (β = 0.28, p < 0.001; η2 = 0.11) when compared to the medium (β = 0.17, p = 0.74; η = 0.04; Z = 2.54, p = 0.005) and high (β = 0.17, p < 0.001; η2 = 0.05; Z = 2.45, p = 0.007) family-affluence groups.

Lastly, the effect of TS on SU also showed a significant change (Z = 2.76, p = 0.003) between the medium FAS group (β = 0.15, p = 0.728; η2 = 0.04) and the high socioeconomic level group (β = 0.05, p = 0.20; η2 = 0.05). In addition, a significant change is observed (Z = 2.18, p = 0.014) between the medium family-affluence group and the group with low family-affluence (β = 0.08, p = 0.139; η2 = 0.01). Therefore, we can also conclude that the model meets the criteria for configurational invariance across socioeconomic groups.

Lastly, Table 6 shows –in addition to the indicators for the general model– the results of testing metric invariance, in which all factor loadings were constrained to be equal for the different groups by sex, age, and socioeconomic status. The fit of the metric invariance models was still good for sex (CFI = 0.923; RMSEA = 0.056; SRMR = 0.046), age (CFI = 0.918; RMSEA = 0.058; SRMR = 0.048), and socioeconomic status (CFI = 0.918; RMSEA = 0.058; SRMR = 0.051), and the increase in CFI was also lower than 0.01 for all models, confirming metric invariance across the samples.

Therefore, despite significant changes in some parameters (e.g. in the indirect effect reported above in the age groups, and the slight variations in the direct effects of family, classmates, and teacher support on SU in the three groups according to socioeconomic level), the goodness of fit indices with parameter constraints (Table 6) confirmed the model’s measurement invariance according to sex, age, and socioeconomic status.

4 Discussion

Research has shown a growing interest in adolescent civic engagement, establishing different measurement indicators that have led to a variety of definitions and instruments used for evaluation. In addition, prior research reports inconsistent results regarding the influence of developmental contexts on youth’s sociopolitical participation, with certain contexts having more importance than others. However, in consonance with the results of this present paper, sense of unity has demonstrated to be essential for analysis of this relationship (Schulz et al., 2010).

The aim of this research was to understand the influence of Spanish adolescents’ socialization contexts on a specific indicator of civic engagement: expected sociopolitical participation. In addition, this study increases awareness about the direct influence of sense of unity on expected sociopolitical participation, as well as indirectly as a mediator in the influence of developmental contexts on sociopolitical participation, and for that matter, on adolescent civic engagement. Moreover, this study adds to the analysis of how the contribution of developmental contexts to sociopolitical participation through the sense of unity differs among boys and girls, among youths in early, mid, and late adolescence, and across different groups according to their family socioeconomic status.

Firstly, our study highlights the important contribution of the immediate developmental contexts for adolescents in establishing sense of unity. These results are consistent with Obst et al. (2002), that found affective, cognitive, and behavioral experiences within the socializing groups which are key for establishing a sense of belonging to these same groups. In addition, this study furthers previous findings by including a new measure –sense of unity–, in a more universal sense than the previously studied sense of community, belonging, or connectedness, all of which refer to more specific feelings about the context in which adolescents participate.

Furthermore, this research provides information about the different contributions of family, friends, teacher, and classmates support to sense of unity. In this regard, peer context –both friends and classmates– showed a stronger contribution to sense of unity than family and teacher, confirming previous findings supporting that quality peer relationships satisfy the need for sense of belonging (Uslu & Gizir, 2017), and are even more influential than family during this developmental stage (Drolet & Arcand, 2013). Along these lines, Huebner and Mancini (2003) reported that friends support in the context of structured activities is essential for group belonging, as well as having a best friend favors integration with other peers. In addition, the relationships that adolescents establish with classmates when they participate in school activities, as well as extracurricular, cultural, or sports activities, contribute to their sense of belonging to the school (Faircloth & Hamm, 2005).

Previous studies on sense of belonging or sense of community have also demonstrated certain family and teacher dimensions to be important for developing a sense of belonging to these contexts. For example, the relationships between teachers and students, the opportunities that educators offer youth to participate in decision making, or the perception of a democratic and safe school climate are key for developing a connection to school (Guillaume et al., 2015; García- Moya, 2020), and a recent metanalysis has shown teacher support to be one of the principal predictors of sense of belonging to school (Allen et al., 2016). On the other hand, the quality of the parent–child relationship and between the parents is essential for developing a sense of belonging to the family (King & Boyd, 2016). However, it must be pointed out that sense of belonging is specific to the context in which the individual develops (Mahar et al., 2013). Therefore, our results showing limited family and teacher influence on sense of unity could be explained because teacher or family influence could be more contextually specific, that is to say, it contributes more to developing a sense of belonging to these specific contexts. Whereas, the peer context seems to be essential in sense of unity –understood as a connection with others beyond the developmental contexts. In this sense, this study highlights the important role of peer socialization on sense of generational belonging –experiences with peers, more than those with family and teachers, are the principal causes of novelties that a generation introduces with regards to the past generation.

In addition, the relationship between sense of unity and expected sociopolitical participation was also significant and relevant. This result is also congruent with other prior works suggesting that shared interests contribute to constructing a sense of community, which in turn fosters social and political participation (Cicognani et al., 2008; Talò et al., 2014).

The present research found that the direct association between support from different developmental contexts on expected sociopolitical participation was not significant. Thus, the most important contribution of this research was demonstrating that the relationships between developmental contexts and youth’s participation in society are established through sense of unity. In this sense, Schulz et al. (2010) indicated that community membership fosters commitment towards it, and therefore, participation for its wellbeing.

In addition, the mediation effects of developmental contexts on expected sociopolitical participation through sense of unity varied depending on the specific context considered, similar to their direct effects on sense of unity. In this regard, the indirect relationship between perceived family and teacher support with sociopolitical participation through sense of unity was weaker than that established with perceived peer support, whether friends or classmates. Thus, higher perceived friends and classmates support was positively associated with higher participation in social and political issues. Socialization contexts –especially that of peers (Wilkenfeld, 2009)– offer an important value towards the adolescents’ commitment, in which sense of unity and group and community membership acquire special relevance for inspiring a proactive attitude, underlining the socializing role of peers in future civic engagement and producing a sense of generational belonging. A possible explanation for this effect is that peers who establish group norms in favor of civic engagement seem to influence the beliefs and ideologies of the other youth (Silva et al., 2004). As Campbell (2008) describes, the sociopolitical participation of many adolescents is related to political ideology influenced by peer alignment.

With regards to the family, despite previous findings showing this context to be a precursor to the youth’s future activities in relationship to society (Kim et al., 2015), and highlighting the idea that an education emphasizing family members’ social responsibility will foster their community participation (Flanagan et al., 1998), our results do not the role of family neither for sense of unity nor expected sociopolitical participation when compared to peer influence. These results are more consistent with other research reporting an unclear relationship between family support and commitment to society (Alesina & Giuliano, 2011). Youth with a strong sense of family belonging tend to help family members first, and once these needs are met they may or may not choose to help others.

The results of teacher support are similar to those of family. The direct relationship established between perceived teacher support and expected sociopolitical participation was insignificant and very weak when taking into consideration the mediator effect of sense of unity. A possible explanation for the limited teacher influence found in this study (compared to peer influence) is that teacher influence occurs precisely through their ability to inspire concern, interests, motivations, etc. in the group of classmates which, as highlighted above, is essential in the socialization of sociopolitical participation at this age.

Regarding the role of sex, our results demonstrate girls to have both higher expected sociopolitical participation and higher sense of unity than boys. The gender gap in civic engagement –in which girls report being more civically engaged and volunteering more often than boys– has been reported in a previous study (Gaby, 2017). In general, girls have a higher level of social engagement than boys, despite boys scoring higher in other participation indicators such as political interest, planning to vote, or following news (Cicognani et al., 2012; Wray-Lake & Shubert, 2019). Our results might be explained by the higher weight of indicators of social engagement versus those of political engagement in the scale used to measure expected sociopolitical participation. In addition, these differences might be explained by gender socialization (Fridkin & Kenney, 2007). Parents tend to promote more autonomy in boys, while being more protective and restrictive with girls, possibly explaining the higher likelihood of girls to participate in organizations oriented towards solidarity and those controlled by adults (Kirchler et al. 1993). Furthermore –following Fox and Lawless (2004)–, families develop different expectations about their son and daughters’ behaviors according to gender roles, considering girls as more passive and compassionate and boys as more oriented towards leadership and public roles. On the other hand, and contrary to our results, previous findings have found that adolescent boys have a stronger sense of community than girls (Chiessi et al., 2010; Cicognani et al., 2012). However, it must be kept in mind that sense of unity refers to a broader concept than sense of belonging to local communities, which is what has typically been measured in previous studies. However, our results show no differences in the influence of perceived support from the contexts in expected sociopolitical participation through sense of unity according to sex. Thus, boys and girls show to benefit similarly from the support from their developmental contexts in their expected political participation through the sense of unity.

The results obtained for each age groups showed no differences in expected sociopolitical participation nor in sense of unity. These results contradict the catastrophic view of some studies which point to a decline in adolescent participation in mid adolescence, caused by a self-centered attitude (Nucci & Turiel, 2009) and more focus on personal problems rather than social concerns (Wray-Lake et al., 2016). Our results are also inconsistent with previous findings indicating that sense of community decreases with age (Albanesi et al., 2007). Model invariance according to age showed that in all age groups the influence of contexts on expected sociopolitical participation was mediated by sense of unity, being higher in the group of peers and classmates than family and teachers. Thus, these results offer a more optimistic view of adolescence, focusing more on the importance of the relationships in their developmental contexts and less on their current developmental stage.

With respect to socioeconomic level, results showed no differences in the effect of family affluence on sense of unity. This result is consistent with those found by Vieno et al. (2005) regarding sense of community, reporting that family affluence was not significant on an individual level. Concerning sociopolitical participation, this study shows significant differences in expected sociopolitical participation according to family affluence –congruent with previous results found by Lenzi et al. (2012). However, the effect size of these difference was negligible. The socioeconomic differences could be caused by other factors beyond family affluence, for example parental education level, in which prior studies have found consistent differences (Metzger et al., 2020; Wray-Lake & Shubert, 2019). Therefore, other socioeconomic dimensions should be explored as possible causes of inequalities in adolescent socioeconomic participation. Nevertheless, this study’s most important contribution was finding that the hypothesized model worked similarly in all three groups according to family affluence. In consonance with the main effect model described by Kawachi and Berkman (2001) and verified in adolescent research (Gecková et al., 2003; MacMillan & Violato, 2008), participation in social networks positively benefits a person’s health and wellbeing regardless of their socioeconomic situation. In this case, support from the different contexts –especially friends and classmates– has shown to influence sociopolitical participation through sense of unity similarly in adolescents from different groups of family affluence.

This study has certain limitations. For example, it is difficult to establish cause-effect relationships between variables due to the transversal design of the research. Likewise, questionnaire methodology has the limitation of self-reported scales. Nonetheless, as previously mentioned in the Method section, all the instruments used or those on which the used scales were based have demonstrated reliability and validity in adolescent populations. Future research complementing this study with measures that collect and contrast data from teachers, family, friends, and classmates could contribute to better comprehension of the model. Another limitation is that education in Spain is only mandatory through 16 years-old and therefore the results for the 17–18-year-old adolescents are not representative of this entire age group but only of those youth that continue within the formal education system. In addition, this study focuses exclusively on the relevance of Bronfenbrenner’s microsystems (Bronfenbrenner, 1979), however future research should also explore the influences of the meso-, exo-, and macrosystems on adolescent sociopolitical participation, as well as the contribution of contexts and sense of unity. It should also be noted that our data was collected in a democratic country and the culture and politics at the moment in which the research was conducted could have affect the results, as Kim and Amna (2015) demonstrated in a previous study. Along these lines, previous comparative studies have found differences in civic engagement related to specific countries’ political systems, and therefore, cultural differences between countries with different welfare systems and policies should be considered. Kim et al. (2015) showed that a country’s level of democratization – expressed through civil liberties and freedom of press– might have an impact on expected sociopolitical participation. In addition, Norris (2002) –in a study conducted in 193 countries– found higher electoral participation and political activism among citizens in countries with more rights and freedoms. Likewise, Karp and Banducci (2008) argued that citizens of incipient democracies are less likely to vote than those in more established democracies. Spain could be considered a country with a solidified democracy and freedom of expression, in which child and adolescent participation in political and social issues has increased, despite differences in adult participation. According to the Youth Report in Spain (INJUVE, 2020), in 2019 Spanish adolescents expressed interest in new social and political topics, such as gender equality, and highlighted their participation in protests and the use of new technologies for personal expression. Following Benedit (2000), youth participation in organizations or associations for community issues has increased in Spain and Europe since the 1990’s. Northern European adolescents report higher levels of participation in politics and associations, however they show lower levels of engagement in other contexts such as family or neighborhood. Conversely, in southern European countries the traditional structures are powerful among citizens and their participation might be more closely related to contexts such as the family. Moreover, changes in family structures and dynamics, as well as the increasing value of individualism and detraditionalization of southern European countries, might influence adolescents’ sociopolitical participation.

Thus, future studies should also conduct cross-national comparisons and longitudinal and periodical analyses of the influence of cultural, political, and national economic contexts of the countries, as well as from their developmental contexts, on adolescent sociopolitical participation.

Amongst the strengths of this research it should be mentioned that our study used a large enough sample for it have significance, as well as an exhaustive and detailed data filter. In a more substantive aspect, the research topic is novel and there is an increasing interest from the scientific community. The majority of research approaches civic engagement from a limited perspective given that it only asks about specific dimensions such as education, politics, etc. (Sherrod et al., 2010). However, this present study adopts a wider perspective from different socialization contexts, and offers a multidimensional focus on sociopolitical participation research. Lastly, the inclusion of sense of community as a key factor for understanding the contribution of the developmental contexts to adolescent sociopolitical participation and analysis of the model’s measurement invariance across sex, age, and socioeconomic status should both be highlighted.

Despite the broad range of measures included in this research, future studies should also analyze other indicators of current sociopolitical participation (i.e., volunteer work, school boards, youth groups, etc.) or other variables such as the duration of participation. Subsequent research should also examine other indicators of civic engagement, such as political interest, political knowledge, intention to vote, or news consumption, and include comparisons using indicators of socioeconomic level, type of school, or family structure, as well as other variables that could explain youth sociopolitical participation.

5 Conclusions

This research analyzed the role of perceived support from different socialization contexts in adolescent development and, more specifically, in their expected future participation in social and political issues, protest, or special interest groups. The most relevant finding is that support from family, friends, teacher, and classmates affects adolescent sociopolitical participation through the sense of unity. Thus, this research highlights the importance of sense of unity for impelling adolescents to action and how the feelings of pertaining to a large group of others promotes actions in the interest of the community.

In addition, this research showed differences in the influence of the developmental contexts on sense of unity, and on the expected sociopolitical participation through it. In this regard, findings indicate the prevalent role that peers –both friends and classmates–, have in fostering a sense of belonging to others compared to family and teachers. Thus, peer experiences are the principal contributors to the sense of generational belonging, which is the most important contributor to adolescents’ expectations of sociopolitical participation. Lastly, this model showed to work similarly for boys and girls, youth in early, mid, and late adolescence, and in adolescents pertaining to families with low, medium, and high material affluence.

Consequentially, educational and social interventions aimed at fostering peer relationships could create bonds promoting awareness about common goals, while simultaneously equipping society with active citizens concerned with issues that affect everyone, something especially desirable in democratic societies. Likewise, schools should not only stimulate positive relationships amongst students but also intensify the student–teacher relationship in order to foster sense of unity. In addition, they should offer educational contexts that are conducive to sociopolitical understanding, include adequate material in course syllabi, and encourage youth to participate in the decisions that affect them. All of this is indispensable for the democratic health of a society.

Lastly, although peers showed to exert the strongest influence on promoting civic engagement during adolescence, one must not forget that the family is still the first socialization context and it is here that the first values are formed. Therefore, family interventions aimed at fostering the sense of belonging amongst members –leading to healthy and beneficial participation for society– must be promoted. That is, if family participation in social and political activities (volunteering, activism, etc.) leads children to act in the interest of their community, then psycho-educational activities should be developed for those families that either do not pertain to groups committed to society, or do not have sufficiently solid enough family ties to pass on their civic engagement to younger generations.