Adolescent subjective wellbeing is a vital aspect in a young individual’s life development. This concept can be affected by a variety of factors such as cultural and developmental contexts, including schools (Casas 2011). Indeed, negative experiences at school, such as becoming a victim of violence, can have both short- and long-term consequences for young individuals (Låftman et al. 2018; Tsaousis 2016; Wolke and Lereya 2015). Conversely, a positive climate at school increases the likelihood of support for victims provided by the school staff (O’Brennan et al. 2014). Although a substantial body of research on adolescent wellbeing already exists (Huebner et al. 2014; Moreira et al. 2015; Žukauskienė 2014) studies that examine the positive impact of school variables upon adolescent wellbeing are scarce. Thus, positive attributes such as a high self-esteem, take-perspective, respect or participation could be responsible for the adolescent wellbeing observed in these privileged settings. Hence, our objective here was to examine if the school climate is a predictor of school violence or victimization in the context of adolescent wellbeing.

1 Adolescent Well-Being

The term wellbeing refers to a personal evaluation of a high quality of life across multiple dimensions (Ben-Arieh et al. 2014). The term may be used to refer to life in general or to restricted specific domains such as friends, school, family, or the community (Huebner 2004). During youth, wellbeing is a critical aspect of promoting a positive development (Casas 2011; Oberle et al. 2011; Proctor et al. 2009). Several studies demonstrate that young individuals with high levels of subjective wellbeing are less likely to engage in misfit behaviors and display better mental and emotional health, academic achievements, and social functioning compared to those with lower levels (Berger et al. 2011; Jiang et al. 2013; Moreira et al. 2015; Seligson et al. 2005).

In Latin America, a growing body of evidence has allowed the development of a multidimensional model of this construct including psychological, subjective, psychosocial, and socio-community well-being (Guzmán et al. 2017; Sarriera and Bedin 2017). In Chile, this theoretical development has been further encouraged by the agenda of the national survey of children’s wellbeing. As an example, the most recent report involved a total of 2572 children and adolescents across three different country districts in order to assess life satisfaction levels (Oyanedel et al. 2014). This research is primarily focused on descriptive analyses. Guzmán et al. (2017) examined differences among subjective well-being measures and found that satisfaction levels were lower in 7th graders versus 5th graders. Another Chilean study performed at a southern city reported that self-esteem and social support were positively related to life satisfaction in a sample of 512 young individuals (San Martín and Barra 2013). These studies have been critical for providing a general view on the dimensions of well-being, however they focused on individual-level associations. On the other hand, the variations in wellbeing levels as a function of the school context and school climate are yet to be described.

Young individuals spend a significant amount of time at school. Therefore, school context has been proposed as a predictor for youth wellbeing (Eccles and Roeser 2008; Roeser et al. 2000). Despite this, the importance of school context is often underestimated and a few studies have examined the effect of school variables, such as school violence and climate upon life satisfaction among youngsters.

2 School Victimization

Estimates from the United Nations Educational, Scientific and Cultural Organization ([UNESCO] 2017) indicate that every year, about 246 million children and adolescents suffer school violence or bullying. School violence refers to aggression and victimization that occurs “inside and outside the classroom, around schools, on the way to and from school, as well as online” (UNESCO 2017, p. 19). In every situation, adolescents can be a victim, an offender, or a witness of school violence. Globally, school violence ranges from bullying to weapon carrying, even homicide (Flannery et al. 2004; Ybarra et al. 2014), however studies confirm bullying as one of the most pervasive forms of school violence.

Bullying is a form of intentional and persistent aggressive behavior carried out by an individual or group against a target commonly characterized by its incapacity for self-defense (Salmivalli 2010). Bullying can be either direct or indirect; the former includes physical (e.g., punching, kicking, hair pulling), and verbal aggression (e.g., malicious rumors, teasing, intimidation). The latter refers to a more relational aggression expressed in several ways (e.g., name-calling, slandering, silent treatment) (Olweus 1993; Richard et al. 2012; Smokowski and Kopasz 2005; Veenstra et al. 2007). On the other hand, boys are more frequently bullying victims compared to girls (Olweus 1993). Male victims report being physically attacked versus female, that suffer indirect bullying more related to gossip and rumors compared to males (Felix and McMahon 2006). Bullying victims report a variety of physical (gastric disturbances, headaches) and psychological symptoms (anxiety, stress) (Fanti and Kimonis 2012; Warren 2011), which negatively affects their academic performance when compared to their peers (Woods and Wolke 2004). Certainly, these adverse experiences have long-lasting negative consequences that could lead to an increased risk for depression and suicide attempts/ideation in the adulthood (Skapinakis et al. 2011). A recent Chilean study by Varela et al. (2018) examined 802 7th graders and found a negative relationship between school climate and school violence and life satisfaction. Further, among adolescents this relationship was mediated by school satisfaction. Importantly, the effect of school climate on the relation between well-being and school violence was examined with a non-hierarchical analysis.

3 School Climate

School climate is a multidimensional concept that can be described as the prevailing atmosphere in an academic institution. School climate sets the norms, values, rules, and structures of an educational institution as a whole (Cohen et al. 2009; Gage et al. 2014). Studies have demonstrated that a positive school climate is associated with reduced bullying reports (Kasen et al. 2004; Orpinas et al. 2003; Waasdorp et al. 2012). Consequently, practice guidelines suggest a special focus on school climate aiming prevention, especially targeting bullying (Hong et al. 2018). Indeed, several school intervention programs aim to decrease teasing and bullying as one of their main goals by changing the school settings. Studies suggest school climate is associated to healthy behaviors, academic self-esteem, and optimism (Bond et al. 2007; Brand et al. 2008; Haynes et al. 1997; Klein et al. 2012; Perkins and Borden 2003). In summary, school climate is a key feature of the school context that provides emotional support for students and increases their wellbeing. The results of the Programme for International Student Assessment (PISA) from 2015 provide an excellent example of this and highlight the importance of the teacher-student relationship for students’ wellbeing. The PISA program found that negative relationships with their teachers had a significant impact on the students’ sense of belonging to the school. Also, students with a higher sense of belonging to school reported higher rates of motivation and achievement (OECD 2017).

As mentioned, school climate is a multidimensional construct, Cohen et al. (2009) developed an extensive review describing four essential dimensions: Safety, teaching and learning, relationships and environmental and structural aspects. More recently, Wang and Degol (2016) also described four domains of school climate including: academic, community, safety and institutional environment. The authors developed a total of thirteen potential dimensions relevant to understand and explain school climate. Hence, despite its confirmed multidimensional nature this study will exclusively focus on two school climate dimensions: interpersonal relationships and bonding with a special emphasis on its relationship with student wellbeing and victimization (Huebner et al. 2014).

Even though several studies have proven the benefits of school climate, Chilean studies have not considered the impact of school climate upon wellbeing at the school-level. Similarly, the protective role of the school climate over school violence, victimization and wellbeing has not been examined. As explained, school climate supports youth wellbeing, protecting them from the negative consequences of school violence. Indeed, Flaspohler et al. (2009) confirmed a negative relationship between being bullied and wellbeing in a sample of 4322 elementary and middle school students. Conversely, negative effects are reduced when students feel the support of their peers and teachers. In Chile, recent studies have described a similar path correlating school climate, school violence and wellbeing, However these studies did not determine whether school climate had a moderating effect or use hierarchical analysis to capture the nested effect of data (Villalobos-Parada et al. 2016).

Our aim here is to examine whether school climate may be a prediction of school violence via subjective wellbeing. Our hypothesis establishes that schools with better school climates will have higher levels of subjective wellbeing reported by the students. We also hypothesize that better school climates will modulate the relationship between school violence and subjective wellbeing.

4 Method

We were interested in exploring whether school climate influences subjective wellbeing and its potential association with school violence. Given the nested nature of our research question (i.e., students nested in schools), Hierarchical Linear Modeling (HLM) is the most appropriate statistical model. HLM permits the partitioning of variance in subjective well-being into its between-school (inter-school) components and its within-school (intra-school) components. Then, the variance in subjective well-being that lies systematically between schools can be modeled as a function of school characteristics (i.e., school climate) while adjusting for characteristics of students.

5 Data Source

5.1 Sample

We use the data of public schools from a single county in the city of Santiago, Chile. Our study included a total of 2013 students (48% female; mean age 14.97 years, SD: 1.86) from 20 schools, 11.9% had a foreign nationality (Peru, Colombia, Dominican Republic, Haiti, Venezuela, Argentina, Bolivia or Ecuador).

5.2 Data Collection

Data were obtained from a single county in the city of Santiago, Chile, as part of the development for a local school climate policy. Data were collected from school community in November 2015 by randomly selecting one grade from each level, from 7th to 12th grade. For this purpose, they use the instrument of the national school climate policy. Self-reported surveys were accrued in student classrooms grouped by grade level during regular class hours under the supervision of external school psychologist to the community using an online website (Survey Monkey). Participants’ information was kept confidential following ethical protocols, participating schools were given informed consent forms.

5.3 Data Analysis

Considering the nested structure of the data we use multilevel analysis approach (hierarchical linear modeling) to examine the multilevel influences on well-being from school variables. Multilevel analysis is ideal for this type of study because it allows us to partition the within-school and between-school variance for our outcome variable (Raudenbush and Bryk 2002). Then, the variance in well-being that lies systematically between schools can be modeled as a function of school characteristics (i.e., interpersonal relationships and bonding) while adjusting for characteristics of students. That is why, by using HLM, it is possible to obtain a better estimates of students outcomes (Arnold 1992).

6 Measures

6.1 Individual Level Variables

6.1.1 Student Well-Being

Our dependent variable was subjective wellbeing, measured by the Brief Multidimensional Students’ Life Satisfaction Scale (BMSLSS) (Seligson et al. 2003), which has been previously validated in Chile (Alfaro et al. 2015). This measure has 6 items and uses an 11-point Likert scale to assess level of satisfaction with different life domains, such as family life, friends, neighborhood, school, self, and life in general (0 = Completely unsatisfied; 10 = completely satisfied). A higher value on the scale indicates greater self-report of well-being. Examples of items are: “How satisfied or dissatisfied are you with ... Your family life; Your friends; Your neighborhood in general; Yourself?”. The scale has an adequate reliability indicator (α = .81) for the current sample. In the analyses, we converted the scale values to z-scores.

6.1.2 Victim of Peer Violence at School

One predictor variable was a measure of peer victimization at school based on the National Chilean Survey on Violence in School Settings (ENVAE by its Spanish initials).This survey was developed in 2006 by the Chilean Ministry of Education and Homeland Office. This scale is composed of 9 items that evaluate the self-report of being a victim of different types of aggressions by peers (e.g., physical, psychological, relational) in the school context, during the school year. It is measured on a 5-point Likert scale, where 1 = “Never “through 5 = “Every day.” Examples of items are “During this year 2014, how often have other students assaulted you in your establishment by “Malicious rumors “, “... physical fights (punches, kicks, hair pulling, etc.) “,“ Teasing or disqualifications.” The scale has a high reliability coefficient (α = .89) for the current sample. A higher value on the scale indicates greater self-report of a peer victimization at school. In analyses, we converted the scale values to z scores.

6.1.3 General Victim of School Violence

A second measure of general school violence victimization was also derived from the National Chilean Survey on Violence in School Settings. This scale is composed of 4 items that assess self-report of being a victim of aggression from different members of the school, such as students, teachers, general staff, and principal, during the school year. It is measured on a 5-point Likert scale, where 1 “Never” and 5 “Every day.” Examples of items are “During the year 2014, how often has someone from the establishment assaulted you? … by a teacher”, “by an Education Assistant (Inspectors, assistant, secretary, etc.)”, “…the Principal.” The scale has an adequate internal reliability (α = .76) for the current sample. A higher value of the scale indicates a greater self-report of a victim of violence at school. In the analyses, we converted the scale values to z scores.

6.1.4 Individual Control Variables

As controls variables, we included age (as a continuous variable), student sex (dummy coded, female = 1), and student nationality (dummy coded, Chilean = 1).

6.2 School Variables

6.2.1 Interpersonal Relationship

Interpersonal Relationship is a composite variable of 5 items that evaluate student perceptions of the quality of interpersonal relationships among school members. This measure is also from the National Chilean Survey on Violence in School Settings. The composite is measured on a 5-point Likert scale, where 1 “Very disagree” and 5 “Very agree.” Examples of items are “How much do you agree or disagree with the following statements about the relationship between students, teachers, and managers of your establishment? ... Students have a good relationship with each other”, “Students and teachers have a relationship of mutual respect”, “Teachers, inspectors, and principal have a good relationship between them.” The scale has good internal reliability (α = .76) for the current sample. A higher value on the scale indicates more positive self-report of interpersonal relationships in the school. In the analyses, we converted the scale values to z scores.

6.2.2 Bonding

A second school climate variable from the National Chilean Survey on Violence in School Settings measures school bonding by asking about feelings of connection and proximity to the school and their members. It uses a 5-point Likert scale, where 1 “Very disagree” And 5 “Very agree”, based on 3 items. The items are “How much do you agree or disagree with the following statements about your establishment?... I feel proud of my establishment”, “I would like to stay at this establishment next year.” The scale has high internal reliability (α = .85) for the current sample. A higher value on the scale indicates more bonding to the school. In the analyses, we converted the scale values to z scores.

7 Results

7.1 Descriptive Results

Table 1 summarizes the descriptive statistics for the studied variables. Table 2 shows the descriptive results comparing the mean score on the BMSLSS with different student-level control variables. We found BMSLSS score differences by age and sex. Older participants reported lower BMSLSS levels compared to younger students (16–21 versus 10–15 year-old). Females reported lower levels compared to males. There were no differences by nationality of the participants.

Table 1 Descriptive statistics study variables
Full size table
Table 2 Descriptive student characteristics by Brief Multidimensional Students’ Life Satisfaction Scale (BMSLSS)
Full size table

Descriptive statistics for all student-level variables (BMSLSS, victim of peer violence and general school violence) relative to school-level variables (interpersonal relationships and bonding) are provided in Table 3. On average, students categorized at higher school levels (i.e., above the mean) in terms of interpersonal relationships displayed higher levels of BMSLSS and lower self-reported victimization. Likewise, students with higher levels of bonding report higher BMSLSS and lower victimization levels.

Table 3 Student descriptive results by interpersonal relationships & bonding
Full size table

7.2 Fully Unconditional Model

We calculated the Intraclass Correlation Coefficient (ICC), which indicates the proportion of the total variance that lies systematically between schools is 7.5%. The lambda reliability estimate of our outcome (BMSLSS) was 0.84.

7.3 Within and Between-School Model

The Intra and Inter-school model results are presented in Table 4. Level 1 accounts for 11.4% of the individual-level variance in BMSLSS in our sample. Victimization by peers was a significant predictor of BMSLSS (β5 = −.47, p < .01), and its relationship with BMSLSS varies significantly among schools. At the school level, interpersonal relationships were significantly associated to average BMSLSS (γ01 = .24, p < .01). All variables were grand-mean centered as level 2 predictors. Finally, bonding was a marginal predictor for the peer victimization/BMSLSS relationship (γ90 = .14, p < .10). As shown in Fig.1 we examined the cross-level interaction with simple slopes; higher school bonding levels (+1 SD over the average) mitigate the peer victimization/BMSLSS association. A 39.4% of the variance in BMSLSS can be explained by level 2 predictors, and 72.3% of the variance in the peer violence victim/BMSLSS association can be explained by level 2 variables.

Table 4 Within-school (Level 1) and between HLM model result
Full size table
Fig. 1
figure 1

Cross-level interaction indicating the relationship between victim of school violence from peers and BMSLSS moderated by school bonding

Full size image

8 Discussion

Consistent with previous reports from both Chilean and international studies (Martin and Huebner 2007; Rigby 2000; Sentenac et al. 2012; Varela et al. 2018), here we provide evidence of a negative association between victimization at school and adolescents’ wellbeing. School climate, and more specifically positive interpersonal relationships have a positive impact upon wellbeing. This is consistent with studies in the U.S. demonstrating that social support in young individuals is associated to wellbeing (Chu et al. 2010). Our results here expand these findings and demonstrate a cross-level interaction of school bonding and peer victimization on wellbeing. A couple of studies have previously assessed the effects of school climate on peer victimization and well-being in Chile and the US (Varela et al. 2018; You et al. 2008, respectively), however, these studies were limited to individual-level analyses. Here we employed a multi-level analysis to confirm the importance of school climate in order to ameliorate the effect of school violence on wellbeing. Further, we found that a school-level indicator such as school bonding mitigates the association between victimization and wellbeing, suggesting that school climate permeates the student body, and therefore the efforts towards a better school climate may also have the potential to improve individual students’ wellbeing. Importantly, this effect was specific to school bonding and we did not find that interpersonal relationships moderated the association between victimization at school and BMSLSS. This result raises important questions on whether the connection to a school might be a more critical indicator for students who are already struggling with their peer relationships, rather than interpersonal connections.

Schools play a crucial role in young individuals’ life and can provide support to their wellbeing, especially via a positive school climate. Indeed, positive school climates decrease the likelihood of school-based bullying (Bandyopadhyay et al. 2009; Klein et al. 2012) and improve peer interactions, academic achievement and social development, diminishing peer rejection (Loukas and Murphy 2007). School violence prevention programs should consider different aspects of the school climate that may limit or decrease the negative impact of violent behavior upon youth wellbeing.

Schools with higher levels of bonding can mitigate the association between peer victimization and reports of life satisfaction. In other words, school communities that promote better social interactions can improve self-reported life satisfaction. This is consistent with other studies that correlate school climate with students’ wellbeing (Flaspohler et al. 2009; OECD 2017), but also suggests school climate is a buffering mechanism and a protective factor against peer victimization.

As described, school climate is associated to life satisfaction and peer victimization, however even in the best possible school climate there will be individuals struggling with social and emotional challenges derived from victimization. Accordingly, Kaufman et al. (2018) emphasize that individual differences should be accounted for, even when anti-bullying prevention programs demonstrate effective, suggesting the need for different levels of support and intervention.

The promotion of school climate policies within schools requires the support of school policymakers granting regular monitoring and the implementation of interventions aiming to improve school climate. Previous studies (Astor et al. 2011; López et al. 2018) emphasize the importance of using a regular system in order to monitor school climate factors; this is also a recommendation from the UNESCO (2017). In particular, school bonding and interpersonal relationships could be key indicators to understand school climate within the school context.

Our results confirm the importance of school life on adolescents’ wellbeing. Indeed, several authors suggest school is a significant proximal context for adolescent development and wellbeing (Eccles and Roeser 2008; Huebner et al. 2014; Tiliouine 2015). Our results also highlight the impact of a positive school experience, and the contribution of relationships among their members to wellbeing. These variables have a direct and protective effect on wellbeing and against peer victimization. Regarding gender differences, our results demonstrate that males reported higher wellbeing levels, consistent with previous reports in Chile (Oyanedel et al. 2014). These differences are evident even after adjusting for this variable in our study.. Despite this, the literature on gender and wellbeing is somewhat controversial and provides mixed results across different cultural contexts (Casas 2011; González-Carrasco et al. 2017), including Chile (Alfaro et al. 2016). For example, Guzmán et al. (2017) found differences in favor of males against females using the Overall Satisfaction Scale (OLS), these results were the opposite when comparing interpersonal relationships for the Domains Satisfaction General Index (DSGI). Therefore, future studies should explore other mechanisms that might explain sex differences for these variables observed among young individuals.

Our study highlights the role of the quality of the teacher-student relationship upon adolescent wellbeing and to prevent bullying. Clearly, peer victimization occurs among students, however teachers can play a significant role in prevention. Our results confirm previous findings demonstrating the impact of the student-educator relationships on adolescent wellbeing (Espelage 2014; Wang et al. 2015).

For several years, school violence research has focused on risk factors at the individual and contextual level, revealing a number of significant predictors for this behavior. Yet, other protective factors at a collective level such as school climate, should be taken into consideration too. The utility of these factors is not limited to prediction but can also inform school psychologists about empirically-grounded prevention initiatives to be implemented at the school. Hence, in order to support youth development we need to focus on both variables. Although our results do not support the role of interpersonal relationships as a moderator variable, they evidence an aspect of school context that supports students’ wellbeing (Huebner et al. 2014). Therefore, interpersonal relationship should be recognized as a meaningful variable across school members. In contrast, peer victimization, is commonly associated to negative peer relationships among students that could have an independent effect upon wellbeing.

Finally, our study had some limitations and faced some challenges that could be solved in future studies. First, our study involved 20 schools, which may not be representative or at least limit the extent of our conclusions. Our study used hierarchical analysis in order to examine the nested effects between participant schools seeking to compensate for this limitation. Secondly, although we used relatively new measures of school climate and school violence, our study shows a Latin American perspective on the potential association of different youth wellbeing predictors. Future studies should incorporate these concepts perhaps as a monitoring system of the students’ quality of school-life. Evidently, we advocate for further research seeking to determine the associations in triad of victimization, school climate, and life satisfaction. This also opens the possibility of future longitudinal studies in order to explore the impact of these phenomena in the development of adolescents.