The effect of teacher-student and student-student relationships on the societal involvement of students

ABSTRACT

The goal of this paper was to examine the relation between teachers and students and between students on societal involvement in Dutch secondary schools. As such, we studied the role of parents on adolescents’ societal involvement and to what extent positive teacher-student and student-student relationships reduced differences in societal involvement due to parental background differences. To estimate this cross-sectional multilevel analyses, a rich combination of datasets from the Netherlands was used, encompassing 4,128 15-year-old students in 58 schools in 2010/2011. The results showed that teacher-student relations and student-student relations were positively associated with societal involvement. The level of societal involvement differed between students’ from households with lower incomes, level of education and employment, even though parenting styles seemed unrelated to societal involvement. Students from higher educated parents were found to benefit more from these positive relationships with teachers. This advantage arguably amplifies the differences in societal involvement between students with lower and higher educated parents. Future studies can give further insight into the role of classroom interrelations using additional longitudinal data or focus on more qualitative observations to explore the role of classroom interrelations and their influence on developing societal involvement.

KEYWORDS:

• Teacher-student relationships
• student-student relationships
• schools
• parents
• peers
• societal involvement

Introduction

In recent years, the interest in the characteristics of effective citizenship education has grown. This interest is fuelled by concerns about the erosion of social coherence, which has inspired many countries to require schools to contribute to the promotion of their students’ citizenship knowledge and skills (Euridyce, 2017). Although little is known about the characteristics of effective citizenship education, research shows the relevance of an open and positive school climate (Geboers et al. 2013; Schulz et al. 2018). Again, little is known about the factors contributing to such a school climate. Citizenship comprises various levels, domains, and aspects of the way people live together. Contribution to ‘the collective good’ is one of these. In many Western countries, responsibilities are shifting from the government and institutions to individual citizens and citizens associations. Citizens should assume more responsibility for their communities, play a more active role in society through voluntary work and rely less on government services. To further the transition from welfare state to a more participatory society, several countries – e.g. the United Kingdom (Kisby 2010), Canada (Fuller, Kershaw, and Pulkingham 2008; Ilcan and Basok 2004) and the Netherlands (Hameleers and Vliegenthart 2016; Tonkens 2014; Verhoeven and Tonkens 2013) – have made an appeal to their residents’ active citizenship.

Growing individualization and social fragmentation among citizens, however, has led to concerns about the degree of active citizenship in societies and its preservation (Blais 2006; Delli Carpini and Keeter 1996; Putnam 2000). Partly against this background, many countries have called upon their schools to contribute to citizenship promotion (Eurydice 2017). One of these countries is the Netherlands, where schools have the obligation to improve their students’ active citizenship and social integration (Dijkstra et al. 2015). As formulated in relevant legislation, schools are expected to teach citizenship to stimulate the capacity and willingness of their students to actively participate in and contribute to society. The underlying idea is that this will improve the students’ civic abilities through the development of knowledge and skills necessary to actively participate in society (Ministerie van Onderwijs en Wetenschappen [Dutch Ministry of Education and Science], 2005). By cultivating positive citizenship attitudes, students are expected to become interested in others and willing to help them, to participate in volunteer work, and to be active in society. In other words, schools are expected to promote citizenship, thus strengthening their students’ involvement in society.

An open and democratic school climate plays an important role in schools realizing the contribution to citizenship promotion that is expected of them. Earlier research shows that such a climate is a promising road towards enhancing the civic competences of students (Geboers et al. 2013; 2015; Isac et al. 2014; Keating and Janmaat 2016; Schulz et al. 2010; Torney-Purta 2002). An open classroom climate provides students with room for discussion and dialogue and nurtures positive interpersonal relationships between teachers and students. In terms of active citizenship, such a climate creates opportunities for students to learn about and practice their (future) role as citizens. Seen from this perspective, schools and classrooms are small communities in which students can learn to engage in social practices and strengthen societal involvement, for example by practicing the necessary skills (e.g. handling differences, cooperating, taking responsibility and making shared decisions). The students’ relationships with others – both other students and their teachers – and the experiences they obtain in both spontaneous and 10.1007anized teaching situations make the school a ‘practice ground’ for participating as citizens. Thus, the nature of the relationships between students and teachers is an important factor for developing as citizens and to become involved in society.

Students, however, differ with regard to their values and social skills. Students from different socioeconomic status (SES) and cultural backgrounds are likely to be socialized differently in their respective families. As a result, the relevance and effectiveness of an open climate may vary (Campbell 2008; Hooghe and Dassonnevillle, 2011; Langton and Jennings 1968; Neundorf, Niemi, and Smets 2016). Campbell (2008) posited two possible effects: Students from a high SES may benefit more from classroom discussions because they are already more accustomed to discussing social issues at home, or students from a low SES may benefit more because they lag behind their high SES peers and have more to gain.

To obtain a better understanding of the mechanisms that may be relevant to the expected effect of education on the promotion of students’ societal involvement, this study focuses on the contribution of the teacher-student and student-student relationships to students’ societal involvement. We will also investigate whether this contribution is the same for all students or whether students from various socioeconomic groups benefit differently. Can schools compensate for a student’s home situation by creating an open classroom climate or do schools sustain or even widen the differences between students from different socioeconomic backgrounds in this respect? As seen below, our investigation of possible differential effects considers differences in the available social, economic, and cultural capital; the influence of parenting styles; and the interaction between these characteristics and teacher-student and student-student relationships on the societal involvement of students.

Theoretical framework

Societal involvement encompasses the perceived relevance or importance of societal issues and topics, often based on a person’s values and interests (Dijkstra et al. 2004; Zaichkowsky 1985, 1994). Contrary to other studies (e.g. Brady, Verba, and Schlozman 1995; Lupia and Philpot 2005; Torney-Purta and Amadeo 2003; Van Deth 2000), our conceptualization of involvement does not imply active behavior. It encompasses an attitude towards societal issues and positively relates to the willingness to participate (Ekman and Amnå 2012; Martin and van Deth 2007). Young people who are willing to learn about societal issues and are willing to participate are more likely to do so if they feel it is necessary (Amnå and Ekman 2014). Since societal involvement is seen as a positive attitude leading to active behavior, it should be understood as a motivator for active behavior (Roberts 2015), and necessary for students to exercise their role as citizens. This entails, for example, that students care about and feel it is important to resolve issues in their community, such as taking care of family members, helping friends, or taking an interest in classmates. It concerns adopting a broader perspective and being engaged in more general issues to benefit society (Ekman and Amnå 2012).

Adolescents might improve their citizenship competences through interaction with other people, through participation and involvement in different social and cultural practices in their daily lives (Lawy and Biesta 2006; Ten Dam et al. 2011). They are active agents, who learn through their own activities, experiences and interactions with others, not passive recipients of what their teachers, peers, and parents do and say (Amnå et al. 2009). To enhance societal involvement, schools need to create an environment in which adolescents are willing and able to interact and participate with other people and are able to reflect on social and cultural practices and decisions (Ten Dam and Volman 2004). Research shows that particularly an open classroom climate, in which students have the opportunity to voice their opinions and are being taken seriously, contributes to their involvement in class (Dijkstra et al. 2015; Geboers et al. 2013; Keating and Janmaat 2016; Torney-Purta 2002). By creating an open classroom climate, in which social issues can be discussed freely from different angles and students can contribute, teachers can increase their students’ willingness, ability, and involvement. By discussing social and societal issues with others (and thus becoming acquainted with various perspectives), by reflecting on their own opinions and by learning to respond to the opinions of others, students practice and experience what it is to be a citizen.

Two factors seem of major importance for creating an open classroom climate: the relationships among students and between teachers and students in the classroom (Campbell 2008; Fraser 1991; Isac et al. 2014; Loukas 2007). Positive relationships among students and between students and teachers are not only positively related to school interest and motivation (Osterman 2000; Furrer, Skinner, and Pitzer 2014; Schunk, Meece, and Pintrich 2014) but to involvement in classroom discussions too (Hamre et al. 2013; Pianta, Hamre, and Allen 2012). Positive relationships among students and between students and teachers make students feel secure and provide room for sharing opinions, which boosts their involvement, motivation and interest in participation (Baker, Grant, and Morlock 2008; Hamre and Pianta 2001).

Students who perceive their teacher as caring, listening, fair and understanding are expected to be more motivated to interact, discuss and participate in class. Positive teacher-student relationships cultivate an environment where students can feel safe and are able, and willing to, participate in discussions (Hamre and Pianta 2006; Midgely, Feldlaufer, and Eccles 1989; Wentzel 2016). This is, students who perceive their teachers as caring, attentive, fair, and understanding are expected to be more motivated to interact, discuss, and participate in class. Participating in these discussions brings them into contact with other ideas and thoughts, which stimulates their willingness to participate -or involvement in- society (Dostie- Goulet 2009; Klofstad 2009; Koskimaa and Rapeli 2015; McIntosh, Hart, and Youniss 2007; Neundorf, Niemi, and Smets 2016).

Positive student-student relationships also can enhance feelings of safety and stimulate the willingness and motivation to participate in class (Baker, Grant, and Morlock 2008; Hamre and Pianta 2001; O’Conner, Dearing, & Collins, 2011; Wentzel 1998) and enhance academic achievement (Roseth, Johnson, and Johnson 2008). If students have better relationships with their peers, there is a greater likelihood that they will feel safe and believe that they are being taken seriously and thus more willing to participate and state their opinions. As such, it is assumed that students with positive relationships with their fellow students are more likely to participate in discussions and dialogues in class, which increases opportunities for promoting involvement in societal issues. The nature of student-student relationships can be different from that of teacher-student relationships. Positive student-student relationships, for instance, do not necessarily have a positive effect on creating an open environment and can be disruptive and less stimulating in class (e.g. Blank and Shavit 2016; Guo, Piasta, Justice, and Kaderavek 2010; Howes 2000; Johnson et al. 1981). At the same time, without these positive relationships with fellow students, expected to stimulate safety in class and the willingness and motivation to participate in discussions, it becomes more difficult to create an open climate. Studying both teacher-student and student-student relationships separately allow understanding the different effects on societal involvement due to the nature of these relationships and to what extent both types of classroom interrelationships stimulate student’s societal involvement.

As noted by Gainous and Martens (2012), student differences may affect whether relationships within the classroom will lead to greater societal involvement (see also Brady, Verba, and Schlozman 1995; Jennings, Stoker, and Bowers 2009; McIntosh, Hart, and Youniss 2007). Family background plays an important role in the development of students’ democratic capacity. Adolescents experience their first social and democratic practices at home (Langton and Jennings 1968; Quintelier 2010), and parents are considered the primary role models of civic behavior (McIntosh, Hart, and Youniss 2007). Differences that occur can partly be attributed to the parents’ socioeconomic status (SES). Parents of higher SES usually possess more social, cultural and economic resources than parents of lower SES, all of which further benefits the social development of their children (Bradley and Corwyn 2002; Brooks-Gunn and Duncan 1997). Children of parents with higher SES often grow up in a more stimulating environment, with more possibilities to develop as citizens, having parents who are better able to transmit the values and skills that contribute to active citizenship (Ichilov 1988; Quintelier, Hooghe, and Badescu 2007).

The following dimensions of a family’s SES are among those relevant: parental education, employment, and income. Highly educated parents, for example, are themselves more involved in society (Pfaff 2009; Jennings and Stoker 2004; Yuen 2013), are more interested in political and societal issues (Verba, Schlozman, and Brady 1995) and engage their children more frequently in discussions (Brady, Verba, and Schlozman 1995; Yuen 2013). Their level of education has a consistent and direct influence on behavioral outcomes, as they create a cognitively stimulating home environment (Bradley et al. 2001). Parents who hold jobs are also regarded as positive role models for their children and have a positive impact on their children’s academic skills and future chance of finding a job (Heinrich 2014; Neal and Hammer 2007). Unemployment, by contrast, is often negatively related to social and academic outcomes for children (Levine 2011; McLoyd 1998; Powdthavee and Vernoit 2013; Sleskova et al. 2006). From this perspective, it can be expected that the relationship between employment and a stimulating and positive environment will also contribute to the societal involvement of students. In addition to these factors, a higher family income provides the material and financial resources necessary to become involved in society and gives parents the opportunities to create a conducive environment and use external resources for enhancing their children’s development also in terms of societal involvement (McLoyd, 1998; Duncan, Morris, and Rodrigues 2011). In sum, in light of the differences between socioeconomic groups, it may be expected that students from higher SES families will be more involved in society than students from lower SES families.

Differences between students from different SES backgrounds may lead to variations in the effect of an open school climate on students (Campbell 2008; Hooghe and Dassonneville 2011; Langton and Jennings 1968; Neundorf, Niemi, and Smets 2016), including some groups of students benefiting more from an open climate with positive student-student and teacher-student relationships than others. As mentioned prior, Campbell (2008) argued that these differences may result in either a compensation effect or an acceleration effect. A compensation effect occurs when students from less-socialized families gain more from an open climate as it compensates for their disadvantages. Students from higher SES families already possess more competences, which means that a climate with positive student-student and teacher-student relationships is less relevant to them. However, empirical research has not provided any clear conclusions in this area. Campbell (2008) found that an open climate might compensate for some SES disadvantages, while other studies were unable to find such an effect (Neundorf, Niemi, and Smets 2016). It is also possible that high SES students benefit more from an open class climate, the reasoning being that students from high SES families already have more competences, so that they can benefit more from the favorable learning opportunities offered by such a climate.

In addition to providing economic, cultural and social resources, parents can also differ with respect to their parenting styles. Previous studies on parenting styles have mostly distinguished between authoritative, authoritarian and permissive styles of parenting. In general, an authoritative parenting style contributes to positive outcomes and establishing social roles for children (Aunola and Nurmi 2005; Yeung, Cheung, Kwok, and Leung 2016). Authoritative parenting is characterized by placing high demands on children and together with an open dialogue about rules and behavior (Steinberg et al. 1992). Such parenting can be distinguished from permissive and authoritarian parenting. Authoritarian parenting is interpreted as only being highly demanding and highly strict, whereas permissive/neglectful parenting lack these attributes and these permissive and authoritarian styles are argued to not maintain a balance of support and demand (Robinson, Mandleco, Olson, Hart, 1995; Skinner, Johnson, and Snyder, 2005). Since parenting styles influence the climate in which the child develops, this raises the question of whether the positive effects of an authoritative parenting style are also felt in the development of societal involvement.

We expect that an open climate, with positive classroom relationships, will contribute to the development of societal involvement in the students, and that with more positive student-student and teacher-student relationships, the students’ societal involvement will be greater. We will also take into account the effect of differences in student background (SES and parenting style) and investigate if and how the classroom climate decreases or increase differences in the societal involvement of students.

Method

Sample

To examine the effect of classroom climate (in terms of student-student and teacher-student relationships) on students’ societal involvement, this study uses data from students in schools for secondary education in the Netherlands. This choice was inspired by the availability of this data, which are well suited to answer the question on which this study focuses. The Dutch situation is in several respects comparable to that in other countries. Dutch schools, like those in many other countries, are expected to pay attention to the promotion of citizenship (Ministerie van Onderwijs en Wetenschappen [Dutch Ministry of Education and Science], 2005). The way in which Dutch students and teachers experience their relationships resembles that of the international average (Schulz et al. 2018). Dutch students have a higher appreciation of their relationships with other students in comparison to students in other countries (Munniksma, Dijkstra, Van der Veen, Ledoux, and Van de Werfhorst, 2017). This means that, if these relationships occur, they should be visible in data collected about Dutch students.

The analyses in this study are based on three combined data sets. Together, they contain a wealth of information collected from robust measurements of the central dependent and independent variables. Our study combined the Cohortonderzoek OnderwijsLoopbanen^5–18 (COOL⁵⁻¹⁸) data with data from the Statistics Netherlands (CBS) and the Inspectorate of Education. The COOL⁵⁻¹⁸ is a large-scale longitudinal panel study intended to collect data on the academic development of students and on the factors that influence this development. It consists of multiple cohorts of students from ages 5 to 18 in the Netherlands. For this study, the COOL⁵⁻¹⁸ 2010/2011 cohort was used, which includes 21,384 Grade 9 students in 149 schools (Zijsling, Keuter, Kuyper, Van Batenburg, and Hemker, 2011). Schools could decide for their students to either participate in this citizenship questionnaire or an English test. Of these students, approximately 40% (8,188 students) completed the module on citizenship competences, which includes a detailed measurement of societal involvement. Using anonymized student ID numbers, COOL⁵⁻¹⁸ data were enriched with data from the Municipal Administration Records of Statistics Netherland behind a remote access environment at the Central Bureau of Statistics (CBS). These records contain unique data on the household, work and income of parents gathered from municipal register data as well as tax-income data. To this data set were added data collected at school level by the Dutch Inspectorate of Education on school denomination and degree of urbanization of the school’s environment.

Little’s missing completely at random (MCAR) test was used to test if the missing data on parenting styles, teacher-student relationships, student-student relationships, and self-efficacy were at least missing at random (MAR), necessary for expectation-maximization imputation (EM). These results indicated the data were indeed MAR (χ2 = .169; df = 25; p = .14), and therefore EM imputation in SPSS was used to impute data. We imputed data for 450 cases for missing data on parenting styles, teacher-student relationships, student-student relationships, and self-efficacy, leading to a total of 4,128 students from 58 schools who were included in the analyses.

Despite a large number of students that were excluded from the analyses, independent sample t-test revealed only minor non-significant differences in the dependent and independent variables between the original and the final sample. Tables A1 and A2 (see appendix) show the frequencies of the standardized continuous variables and categorical data used in the analyses for both student-level and school-level.

Variables

Dependent variable: societal involvement

Societal involvement was based on a Likert scale of nine items derived from the Citizenship Questionnaire (Ten Dam et al. 2011), which was an integral component of the COOL^5–18 student questionnaire. Example items included the following statements: ‘People should listen to each other, even if they disagree,’ and ‘It’s important to learn about other cultures.’ These questions encompass a personal importance and affection toward issues in society vital for understanding societal involvement. Students rated these statements from 1 (‘does not apply to me at all’) to 4 (‘totally applies to me’). Cronbach’s alpha of the scale was .85.

Independent variables (school level)

Teacher-student relationships

This scale is based on items of a questionnaire developed by Peetsma, Wagenaar, and de Kat (2001) to investigate how students perceive their relationship with their teachers. Example items: ‘Teachers know how I feel’, ‘I feel comfortable with teachers’ and ‘I have good contact with teachers’. These were measured using a 5-point Likert scale (7 items), ranging from 1 (‘does not apply at all’) to 5 (‘applies perfectly’). One negatively formulated question was recoded. Cronbach’s alpha of the scale is .82.

Student-student relationships

This scale is based on items of the COOL⁵⁻¹⁸ student questionnaire asking students how they perceive the relationship with peers in their class. Example items: ‘I have a lot of contact with classmates’ and ‘We have a great class’. The six items were measured using a 5-point Likert scale, ranging from 1 (‘does not apply at all’) to 5 (‘applies perfectly’). Two negatively formulated questions were recoded positively. Cronbach’s alpha of the scale is .84.

Independent variables (student level)

Income

Parental income is the sum of both parents’ nett income, based on CBS data collected from the Municipal Administration and the Tax Office in 2010. To account for the skewness of the distribution and to have approximately equal respondents in the different income groups income was recoded as 0 (‘lower income’), 1 (‘medium income’) or 2 (‘higher income’).

Parental education

Parental education was measured by asking parents the education level of themselves and their partner. Parental educational level is based on the highest level of education attained within the family and were coded as 0 (‘lower’), 1 (‘medium’) and 2 (‘higher’). Lower education included all parents with no or low vocational diplomas, higher educated parents all parents with pre-university education and higher and middle everything in between.

Parental occupational status

Parental occupational status is based on the 2010 CBS Municipal Administration data. Occupation status was coded for both father and mother as (1) unemployed, (2) employee, (3) self-employed, (4) retired and (5) other. These data were recoded as 0 (‘both parents not working: either unemployed or retired’) 1 = (‘one parent working’: one parent either employee or self-employed and the other unemployed or retired)’); and 2 (‘both parents employed or self-employed’). Parental occupations coded as ‘other’ were excluded from the analyses due to their very small numbers.

Parenting style

Parenting style was based on four items from the COOL^5–18 parent questionnaire. Parents indicated whether their child had a say in the following decisions: (1) at what time their child had to be home, (2) what television programs their child was allowed to watch, (3) when their child does homework, and (4) for how long their child is allowed to use the Internet or play video games. For each item, parents could choose between the following categories: 0, parents decide without checking with their child; 1, parents decide after discussing things with their child; 2, parents and child together make the decision; 3, the child decides after discussion with his or her parents; and 4, the child decides without discussion with his or her parents. For each item, the responses were categorized as follows: 0, authoritarian; 1, authoritative; and 2, permissive.

Control variables: student level

Gender

Boys were coded as 0 and girls as 1.

Academic efficacy

The students’ academic efficacy was measured using a scale based on the Patterns of Adaptive Learning Scales (PALS, Midgley et al. 2000). Students were asked to indicate whether they were confident that they were competent to do their class work on a 5-point Likert scale ranging from 1 = ‘does not apply at all’ to 5 = ‘applies completely’. Example items: ‘I’m certain that I can master the skills taught in class this year’ and ‘I can do almost all the work in class if I don’t give up’. Cronbach’s alpha of the scale is .84.

Control variables: school level

School denomination

The denomination of the school is based on data from the Inspectorate of Education, and coded as 0 = ‘public schools’ and 1 = ‘private government-funded schools’. Most private government-funded schools are Catholic or Protestant schools, all of which receive the same government funding as public schools.

Urbanization

The degree of urbanization is based on the classification used by the Dutch Inspectorate of Education, coded as 0 = ‘low urbanization’, 1 = ‘average urbanization’ and 2 = ‘high urbanization’.

Data analysis

A stepwise multilevel analysis was used to examine the effect of school and student background variables on the students’ societal involvement on student and school level. The models were estimated using SPSS 24 with HLM methods and maximum likelihood estimates for societal involvement. Six separate models were estimated step by step in which variables were added in each consecutive model. First, to determine the necessity for multilevel analyses, the intraclass correlation coefficient (ICC) was estimated in the null model (model 0). In model 1, gender (boy/girl), ethnicity (native/non-native) and self-efficacy as a personal factor were added as control factors. In model 2, the parental factors income, occupational status, parental education, and parenting styles were added. In model 3, to examine the association between an open climate and societal involvement, teacher-student and student-student relationships were added to this model. In the fourth model, the interaction effects of parental education and student-student and teacher-student relationships were estimated (model 4). Finally, to examine and account for possible differences between schools, the student-student and teacher-student relationships were aggregated at a school level, and school denomination and urbanization were added to the analyses to control for school differences (model 5). For each model, both the proportional reduction in total variance $R_1^2$and the residual variance of the group means $R_2^2$ were estimated (Snijders and Bosker, 2012). The log likelihood model estimation was used to determine whether adding variables improved the model, with a chi-square test to assess whether the differences between models were statistically significant. Effect sizes were calculated by standardizing the estimates.

Results

Table 1 shows the six estimated models for student societal involvement. Adding variables to the models results in an increase in proportional variance explained. Each of the Models 0–5 explains more variance than the baseline model at a statistically significant level. However, a comparison of the subsequent models reveals that Models 4 and 5 do not add substantially to Model 3. Since the hypothesized interaction effects were modeled in Model 4, these models remained in the analysis.

Table 1. Standardized effects of multilevel regression on societal involvement.

SOCIETAL INVOLVEMENT	VARIABLES	M0B(SE)	M1B(SE)	M2B(SE)	M3B(SE)	M4B(SE)	M5B(SE)
	Constant	.037 (.045)	−.172 (.041)	−.101 (.153)	−.097 (.151)	−.093 (.151)	−.095 (.158)
LEVEL 1	Gender student (ref = boy)
CONTROL	Girl		.309 (.029)	.302 (.029)	.302 (.029)	.301 (.029)	.299 (.029)
	Generational background (ref = native)
	Non-native		.320 (.048)	.317 (.050)	.337 (.049)	.335 (.049)	.320 (.049)
	Self-efficacy		.205 (.015)	.194 (.015)	.162 (.015)	.163 (.015)	.161 (.015)
PARENTS	Occupation (ref = no work)
	One parent works			−.184 (.102)	−.171 (.100)	−.167 (.101)	−.157 (.101)
	Both parents work			−.250 (.089)	−.240 (.088)	−.242 (.088)	−.230 (.088)
	Income (ref = lower)
	Middle			.013 (.038)	.015 (.037)	.016 (.037)	.015 (.037)
	High			.079 (.045)	.093 (.045)	.095 (.045)	.095 (.045)
	Education parents (ref = lower)
	Middle education			.010 (.041)	.010 (.041)	.008 (.041)	.009 (.041)
	Higher education			.169 (.044)	.182 (.044)	.181 (.044)	.177 (.043)
	Parenting: time home (ref = authoritarian)
	Permissive			.044 (.358)	.072 (.353)	.075 (.353)	.063 (.353)
	Authoritative			.110 (.060)	.101 (.060)	.101 (.059)	.104 (.059)
	Parenting: watch television programs (ref = authoritarian)
	Permissive			−.199 (.093)	−.178 (.091)	−.167 (.091)	−.174 (.091)
	Authoritative			−.032 (.089)	−.021 (.088)	−.016 (.088)	−.015 (.088)
	Parenting: homework autonomy (ref = authoritarian)
	Permissive			.008 (.104)	−.030 (.103)	−.035 (.102)	−.039 (.103)
	Authoritative			−.090 (.103)	−.108 (.102)	−.113 (.102)	−.116 (.102)
	Parenting: Internet/video games (ref = authoritarian)
	Permissive			.096 (.070)	.089 (.069)	.091 (.069)	.091 (.069)
	Authoritative			.085 (.063)	.077 (.062)	.076 (.062)	.075 (.062)
RELATIONSHIPS	Student-student relationships				.024 (.015)	.065 (.031)	.062 (.031)
	Teacher-student relationships				.143 (.015)	.089 (.030)	.089 (.030)
INTERACTION	Student-student relationships * education parents middle					−.041 (.045)	−.041 (.039)
	Student-student * education parents higher					−.063 (.044)	−.063 (.039)
	Teacher-student relationships * education parents middle					.039 (.044)	.040 (.038)
	Teacher-student relationships * education parents higher					.101 (.043)	.100 (.037)
SCHOOL LEVEL	School student-student relationships						.272 (.133)
	School teacher-student relationships						.098 (.118)
	Urbanization (ref = low)
	Average urbanization						−.009 (.061)
	High urbanization						.290 (.085)
	Denomination (ref = non-denominated)
	Denominated						−.077 (.060)
	ICC	9%
	$R_1^2$		9%	12%	14%	14%	16%
	$R_2^2$		7%	8%	18%	18%	18%

Parameters printed bold indicate significant effects with p < .05.

As Table 1 reveals, more positive relationships with teachers were associated with higher levels of involvement in society (.089). Furthermore, students from higher-educated parents who experience a positive teacher-student relationship are relatively more involved in society than students from less educated parents (.101). Moreover, concerning student-student relationships, both individual perceptions (.062) and average perceptions on the school level (.272) had a positive effect on students’ societal involvement. No significant interaction effect was found between student-student relationship and level of parental education.

In general, students of higher-educated parents were more involved in society than those from less educated parents (.177). Furthermore, students with high-income parents were more involved in society than those with low-income parents (.095). Students with both parents working were less involved in society than students with unemployed parents (−.230). Parenting style was unrelated to societal involvement.¹

Finally, girls on average were more involved than boys (.299), and non-native students were more involved than native Dutch students (.320). Moreover, academically effective students were more involved in society than those with lower academic self-efficacy (.161). Because academic self-efficacy can, to a certain extent, also be regarded as an indication of academic competence, this finding indicates that students who are more confident about their level of competence are also more involved in society than their less competent peers. The findings also indicate that students in schools located in dense, urban areas are on average more involved in society than those in schools of less dense areas (.290).

Conclusions

This study investigated the extent to which positive student-student and teacher-student relationships contribute to the development of societal involvement in students. Several Western countries developed policies that expect schools to promote students’ societal involvement as an element of citizenship promotion and socialization, especially with growing concerns about eroding social cohesion and an ongoing shift from a welfare state toward a more participatory society. Societal involvement is considered an important condition for becoming an active and responsible citizen who contributes to his or her community and to society as a whole. The increased attention being paid to socialization raises questions, for example about the factors contributing to its effectiveness. Although an open classroom climate seems to be one of these factors (Geboers et al. 2013; Schulz et al. 2018), we still know little about factors contributing to such a climate. This study focuses on the contribution that could be made by the quality of the student-student and teacher-student relationships, more in particular their effect on one aspect of citizenship: the societal involvement of students.

Our study indicated that the classroom and school climate – more specifically a positive relationship between the teacher and his or her students – can foster the students’ societal involvement. Our findings support the idea that students who perceive their teachers as caring, understanding, and listening are better able and more willing to engage in classroom activities. It is argued that when students feel safer in their classrooms, they are likely to become more participate in class (Furrer, Skinner, and Pitzer 2014; Hamre et al. 2013; Hamre and Pianta 2001; Wentzel and Brophy 2014). Lawy and Biesta (2006) argued that as students discuss democratic and social practices and become acquainted with them, they will grow into their role as citizens and their societal involvement increases. International comparative research has shown that students who experience the classroom climate as open also achieve higher citizenship competence scores (Schulz et al. 2018). Our findings support the existence of an accelerating effect of parental education on the relation between positive teacher-student relationships and societal involvement. Although all students gain from having a supportive and caring relationship with their teachers, this relationship is stronger for students with highly educated parents than for those with less educated parents. In line with the hypothesized acceleration effect proposed by Campbell (2008), this finding indicates that students from more advantageous backgrounds are more able to benefit from an open climate as they are more experienced and better equipped to partake in discussions on societal issues.

The study reveals that parental resources (indicated by parental level of education, employment, and income) are relevant predictors for student societal involvement. We have simultaneously estimated the effects of parent SES and found that financial resources (as based on official tax data) are positively related to student societal involvement. Moreover, students from highly educated parents are more involved in society than students from less well-educated parents. A plausible explanation for these findings is that well-educated educated parents are often more interested in discussions about political and societal issues and engage in them more frequently (e.g. Verba, Schlozman, and Brady 1995; Jennings and Stoker 2004; Yuen 2013), which would imply that students from these families grow up in an environment that is more conducive to becoming involved in society.

Contrary to our expectations, the results also showed that children with unemployed parents are more involved in society than children from families in which both parents are employed. In general, parental unemployment is negatively related to their children’s social and academic outcomes (e.g. Levine 2011; Powdthavee and Vernoit 2013; Sleskova et al. 2006). Several reasons may have accounted for these findings. First, these effects were measured simultaneously, and focusing on the effects of parental education and income leaves only a positive effect for employment.² That is, unemployed parents, when controlled for income and parental education, are more involved and expected to have more time to participate in society – for example, in their neighborhood – and thus serve as role models for their children. As the three aspects of SES are correlated, the reported effect may be influenced by a few atypical cases – particularly since the number of students with unemployed parents is rather low (see Table A3 in the Appendix). Finally, a speculative explanation for the negative relation between parental social status and their children’s societal involvement is the environment in which these students grow up, such as how social problems impact their families’ everyday lives, motivates them to become involved.

Furthermore, a positive relationship between peer relationships and student societal involvement was found, which indicates that students with a more positive orientation towards their fellow students tend to be more involved in society. Moreover, at school level, we found an even stronger relationship between peer relationships and societal involvement. This means that when students in a particular class generally grade their relationships with their peers as good, they are more inclined to be involved in society. This finding speaks to the necessity of student-student relationships for active involvement, security, and motivation in class (Johnson 1981; Rubin, Bukowski, and Parker 1998; Slee and Skrzypiec 2016). It should be noted that the analyses were based on cross-sectional data, which do not allow for causal statements. A longitudinal design or more qualitative data are required to gain a better understanding of the mechanisms at play.

We were unable to find an association between parenting style and societal involvement. It should be noted here that the categories were unequally distributed – most parents indicated adopting an authoritative parenting style and lower numbers reported authoritarian parenting and permissive parenting. Even though these subjects of parenting and parenting styles informed democratic processes and decision making at home, future studies should explore additional parenting subjects, possibly closer to direct societal issues that might stimulate students to discuss these issues and in turn, according to our argumentation, lead to greater societal involvement.

To conclude, this article shows that both the relationship between students and teachers and students’ relationships with their peers play important roles in stimulating students to become involved in society. These relationships within the school confirm our supposition that schools can contribute to societal involvement, which arguably motivates students to participate in society. The association found between school, student background, and student societal involvement adds weight to schools’ mission to enhance active citizenship and highlights the importance of further research to understand the mechanisms involved to increase our knowledge of effective schooling and school improvement in the citizenship domain.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by [NWO Programming Council for Educational Research (PROO)] ‘[411-12-037]’

Notes on contributors

Frank H. K. Wanders

Frank H. K. Wanders MSC. is a postdoctoral researcher in Sociology at the University of Amsterdam and ProDemos. His research interests are mainly interdisciplinary quantitative research on the political socialization of adolescents.

Anne Bert Dijkstra

Prof. Dr. Anne Bert Dijkstra is a professor of Supervision and effects of Socialization in Education at the University of Amsterdam, and Program Director at the Dutch Inspectorate of Education. His academic research focuses on social outcomes of education.

Ralf Maslowski

Dr. Ralf Maslowski is an assistant professor in Educational Sciences at the University of Groningen and researcher at the Netherlands Institute for Social Research. His research interests include educational policy, educational administration, comparative education, citizenship education and the internationalization of schools.

Ineke van der Veen

Dr. Ineke van der Veen is a senior researcher at the Kohnstamm Institute of the University of Amsterdam. Her research interests include the socio-emotional development of children and adolescents from different ethnic and socio-economic backgrounds and factors influencing this development and citizenship education

Notes

1. Correlations between the variables are presented in table C (see Appendix).

2. In the analyses, the effects are estimated simultaneously, thus accounting for the other aspects of socioeconomic status. Since the three aspects of socioeconomic status are correlated, the reported effect may be influenced by a few atypical cases – particularly as the number of students with unemployed parents is rather low (see Table A1, Appendix).

Appendix

Table A1. Frequencies continuous variables.

Variable	Mean	SD	Min	Max
Dependent
Societal involvement	0	1	−3.071	2.642
Level 1 variables
Student-student relationships	0	1	−5.220	1.853
Teacher-student relationships	0	1	−4.144	3.494
Self-efficacy	0	1	−4.353	2.580
Level 2 variables
School student-student relationships	0	1	−.545	1.028
School teacher-student relationships	0	1	−.793	1.257

Table A2. frequencies of categorical variables per student.

	N	%
Gender
Male	2,099	50.2%
Female	2,084	49.8%
Generational background
Native	3,673	87.8%
Non-native	510	12.2%
Occupation
Neither parent works	127	3%
One parent works	272	6.5%
Both parents work	3784	90.5%
Income
Low	1,309	31.3%
Middle	1,703	40.7%
High	1,171	28%
Parents education
Low	1,814	43.3%
Medium	1,162	27.8%
High	1,207	28.9%
Parenting: time at home
Authoritarian	263	6.2%
Permissive	7	0.01%
Authoritative	3,913	93.4%
Parenting: television program
Authoritarian	131	3.1%
Permissive	1,351	32.3%
Authoritative	2,701	64.6%
Parenting: homework
Authoritarian	104	2.5%
Permissive	2,202	53.6%
Authoritative	1,877	44.9%
Parenting: Internet/video games
Authoritarian	172	4.1%
Permissive	1,109	26.5%
Authoritative	2,902	69.4%
Urbanization
Low	1,595	38.1%
Average	2,099	50.2%
High	489	11.7%
Denomination
Non-denomination	1,637	39.1%
Denomination	2,546	60.9%

Table A3. Correlations.

Variable	1	2	3	4	5	6	7	8	9
1: Societal involvement	1	.18	.09	.21	.06	.05	−.04	.04	.01
2: teacher-student relationships	.18	1	.24	.21	−.03	.03	−.02	.06	.01
3: student-student relationships	.09	.24	1	.10	.01	−.01	−.03	.04	−.03
4: efficacy	.21	.21	.10	1	.03	.03	−.03	.01	−.04
5: income	.06	−.03	.01	.03	1	−.03	.07	.08	.06
6: parenting: time at home	.05	.03	−.01	.03	−.03	1	.09	.08	.17
7: parenting: television program	−04	−.02	−.03	−.03	.07	.09	1	.40	.42
8: parenting: homework	.04	.06	.04	.01	.08	.08	.40	1	.47
9: parenting: Internet/video games	.01	.01	−.03	−.04	.06	.17	.42	.47	1