Academic competence, teacher–student relationship, and violence and victimisation in adolescents: The classroom climate as a mediator

International Journal of Environmental Research and Public Health Article Academic Competence, Teacher–Student Relationship, and Violence and Victimisation in Adolescents: The Classroom Climate as a Mediator Teresa I. Jiménez 1, David Moreno-Ruiz 2,* , Estefanía Estévez 3, Juan Evaristo Callejas-Jerónimo 4, Ginesa López-Crespo 1and Sonsoles Valdivia-Salas 1   Citation: Jiménez, T.I; Moreno-Ruiz, D.; Estévez, E.a; Callejas-Jerónimo, J.E.; López-Crespo, G.; Valdivia-Salas, S. Academic Competence, Teacher–Student Relationship, and Violence and Victimisation in Adolescents: The Classroom Climate as a Mediator. Int. J. Environ. Res. Public Health 2021,18, 1163. https://doi.org/10.3390/ ijerph18031163 Academic Editor: Paul B. Tchounwou Received: 23 December 2020 Accepted: 25 January 2021 Published: 28 January 2021 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). 1Department of Psychology and Sociology, University of Zaragoza, 44003 Teruel, Spain; [email protected] (T.I.J.); [email protected] (G.L.-C.); [email protected] (S.V.-S.) 2Department of Social Psychology, University of Valencia, 46010 Valencia, Spain 3Department of Health Psychology, Miguel Hernández University, 03202 Alicante, Spain; [email protected] 4Department of Applied Economics I, University of Sevilla, 41018 Sevilla, Spain; [email protected] *Correspondence: [email protected]; Tel.: +34-96-382-81-21 Abstract: School violence is a serious social and public health problem prevalent worldwide. Although the relevance of teacher and classroom factors is well established in the literature, few studies have focused on the role of teacher perceptions in school violence and victimisation and the potential mediational role of classroom climate in this relationship. A total of 2399 adolescents (50% girls), aged between 11 and 18 years (M = 14.65, SD = 1.78) and enrolled in five Spanish Secondary Compulsory Education schools completed measures of classroom climate, school violence towards peers and perception of peer victimisation, and their teachers informed about their academic competence and the teacher–student relationship. Correlational analyses revealed that whereas academic competence perceived by the teacher was negatively related to overt violence and victimisation, its relationship with pure relational violence was positive. Structural equation modelling analyses showed that variables of classroom climate (involvement, affiliation, and teacher support) perceived by the students functioned as partial mediators between teacher perceptions of academic competence and of teacher–student relationship and violence and victimisation. In the mediational model, teacher perception of academic competence acted as a direct protective factor against violence and victimisation, and teacher perception of teacher–student relationship acted as a direct risk for violence, as well as an indirect protective factor through classroom climate for victimisation. The interpretation of these results points to the importance of the teacher’s subjective perceptions in the prevention of violence and victimisation problems and their practical implications for the classroom climate perceived by students. Keywords: school violence; school victimisation; classroom climate; academic competence; teacher– student relationship; adolescents; mediation 1. Introduction School violence is a serious social and public health problem prevalent worldwide [1–5]. This problem greatly worries parents [ 6 ], teachers [ 7 ], researchers [ 8 ], and clinicians [ 9 ] due to the significant and negative—sometimes devastating—consequences for the physical and psychological health of the students involved [ 10 – 12 ], with long-term repercussions in the person’s integral development [ 13 , 14 ]. Among the different acts comprised in the term of school violence (i.e., aggression towards school staff, property damage, vandalism on the campus, bullying), the current study focused on violence perpetration towards peers, that is, on nonaccidental behaviours aimed at other classmates to cause harm intentionally; and on peer victimisation, that is, the perception of being exposed to negative actions by one or more students with the intention of hurting the victim [ 15 ]. These acts Int. J. Environ. Res. Public Health 2021,18, 1163. https://doi.org/10.3390/ijerph18031163 https://www.mdpi.com/journal/ijerph Int. J. Environ. Res. Public Health 2021,18, 1163 2 of 16 can manifest as physical (pushing, hitting, kicking, shoving, etc.), verbal (name-calling, taunting, threatening, etc.) or relational violence (spreading false rumours or malicious gossip, exclusion from activities, or withdrawal of friendships). The first two forms of violence–victimisation have been labelled as “direct or overt” bullying, and the third as “indirect or relational” bullying. To understand the relevant negative consequences of these behaviours (for a review see [ 16 ]), it is important to note that the perpetrator usually acts with the desire and intention to dominate and to exert control over the other person [ 17 ], creating a dynamic of violence from which the victims normally cannot defend themselves or escape by their own means, due to the imbalance of power between the people involved [18,19]. Considering prevalence, international data indicate great variation among the studies carried out in different countries. For example, in a study conducted with adolescents from 43 countries (Europe, USA, and Canada) by the World Health Organization (WHO) [ 20 ], the rates of bullying perpetration varied from 1 to 36%, and the rates of bullying victimisation between 2 and 32%. Hymel and Swearer [ 21 ] argued that the variation rate depends on the country of origin of the sample and the way that school violence was assessed. In adolescent Spanish samples, García-García et al. [ 22 ] conducted a systematic review of the prevalence of school violence in 32 studies, which yielded a rate of 11.45% for overall bullying. Moreover, in cross-cultural surveys, rates of violence perpetration and victimisation are consistently higher for boys than for girls [ 23 ]. More specifically, whereas initial research found that boys perpetrate and experience more overt violence, and that indirect forms of violence may be more typical of girls [ 24 , 25 ], later evidence suggests that both sexes are involved in similar levels of relational violence and victimisation [26,27]. In the search of causes for school violence and victimisation, many authors have explored the personal profiles of aggressors and victims and the characteristics of their family, school, and community environments (for a review from an ecological perspective, see [ 28 ]). Specifically, within the school context, most studies have focused on the student’s point of view by analyzing the perception of low teacher and peer support [ 12 , 29 , 30 ] or the importance of group processes such as the search for a social reputation and peer acceptance [28,31]. However, few studies have focused on the role that teachers’ perceptions of their students can play in the explanation of school violence and victimisation. Specifically, in this study, we focus on academic competence and the teacher–student relationship perceived by the teacher. Concerning academic competence, previous studies in the field have shown a small but significant negative correlation between peer victimisation and academic achievement across studies (for a meta-analytic review, see [ 32 ]). In addition, a meta-analysis performed by Savage et al. [ 33 ] informs of the consistency of a negative and moderate relationship between academic achievement and violent antisocial behaviour. More specifically, traditional studies on bullying victimisation [ 15 ] have viewed aggression towards peers as a reaction to frustrations and failures at school. However, later studies provided no evidence for this hypothesis. For example, Woods and Wolke [ 34 ] found no relationship between academic achievement and direct forms of violence perpetration, and a positive and significant relationship with indirect or relational violence perpetration. However, most of these studies used academic achievement (i.e., grade point averages) as an objective indicator of academic competence and few of them have incorporated other teacher ratings of academic functioning, such as subjective ratings of effort and interest in classroom activities [ 32 ]. In this study, we are interested in extending previous literature in this field, considering a broader definition of academic competence because it has been shown that teachers’ subjective expectations about the performance of their students affect their interactions with them, influencing the classroom climate as a whole [ 35 ]. Moreover, it is well established that the quality of school and classroom climate is related to general antisocial behaviour [ 36 ] and, specifically, to violence and victimisation problems [11,12]. Concerning the teacher student–relationship, Bouchard and Smith [ 37 ] have recently argued that researchers and educators should recognize more fully the role of teacher– Int. J. Environ. Res. Public Health 2021,18, 1163 3 of 16 student relationship quality in students’ bullying–victimisation experiences. Drawing from the social–ecological system theory, these authors claimed the need to consider mesosystem analyses—such us teacher–student relationship quality—to deeply understand school violence and victimisation experiences. Specifically, some authors have argued that peer interactions in classrooms, including violence and victimisation, are implicitly shaped by the quality of teacher–student relationships because teacher–student interactions can be seen as a relational model [ 37 – 39 ]. Indeed, previous empirical evidence attests to the importance of teacher–student relationships in psychosocial development (for a review, see [ 40 ]). For example, students who report better relationships with teachers show higher self-esteem and prosocial behaviour, and fewer depressive and problem behaviours such as peer aggression [ 41 , 42 ]. Conversely, other studies have shown that teacher–student relationships that are distant and conflictive contribute to higher levels of school violence and victimisation [ 43 , 44 ]. Research has also revealed that low perceived support from teachers or perception of low teacher involvement is related to high school aggression [28,41,45]. However, most of the cited literature about the benefits of teacher–student relationships on psychosocial adjustment has focused on primary school and few studies have analysed these benefits in secondary schools, where the school structure does not allow teachers to have as many opportunities to develop significant and caring relationships with their students. Moreover, the existing literature has almost exclusively considered teacher–student relationships from the students’ point of view. This study aims to contribute to filling this gap by analysing the link between the teacher–student relationship perceived by the teacher and violence and victimisation in secondary school. As noted before, teachers are one of the most influential agents in establishing the quality of classroom climate [ 46 , 47 ]. That is, teachers who engage in their practice and show affective attunement and support toward their students could contribute to generating a classroom climate where the students perceive support from others, participate in proposed activities, and view peers as friends [ 48 ]. Therefore, we can expect that the teacher’s factors have an indirect influence on the levels of violence and victimisation through their contribution to the classroom climate in which peers interact with each other. Indeed, some previous but scarce studies support the idea that classroom climate could serve as a mediator between teacher practices and peer violence [ 49 ]. Considering teacher perceptions, the original research conducted by Harris and Rosenthal [ 50 ] showed that teacher expectations about their students were related to academic outcomes through a classroom factor such as a warm and affective climate more than through the quality of dyadic interactions. Subsequent studies [ 35 , 51 ] have shown that teachers with high expectations for their students created a different instructional and socioemotional environment, characterized by a better structuring of the teaching process, more feedback, a greater number of open questions and more positive and preventive management of disruptive behaviour in the classroom. However, most of the previous literature focused on teacher perceptions and classroom climate have focused on academic outcomes and not on behavioural outcomes such as school violence and victimisation. In sum, considering that (1) teacher perceptions of academic competence and of teacher–student relationship are related to school violence and victimisation, (2) teacher perceptions about their students affect classroom climate, and (3) classroom climate is related to violence and victimisation among peers, we might hypothesize that classroom climate could be a mechanism linking teacher perceptions and violence and victimisation. We have not found any previous studies in the field jointly analysing these factors, still less focusing on the teachers’ point of view. The Present Study In accordance with the previous literature, in the present study, we seek to (1) explore the relationships between the study variables (i.e., academic competence and teacher– student relationship perceived by teachers, classroom climate, school violence, and school victimisation); (2) determine whether classroom climate mediates the relationship be- Int. J. Environ. Res. Public Health 2021,18, 1163 4 of 16 tween teacher perceptions of academic competence and of teacher–student relationship, and school violence and victimisation; and (3) examine the role of sex as a potential moderator of these relationships in a sample of Spanish adolescents. We expect to find the mediating role of classroom climate and no sex differences in the pattern of mediational relationships among the study variables. 2. Materials and Methods 2.1. Procedure The data we are analysing in the present study were collected as part of a larger study on violent behaviour in Spanish adolescents. Initially, a letter with a summary of the research project was sent to the principal of each school to explain the purpose of the research and to request permission to carry out the study. The management team of each school requested the collaboration of their teacher staff. Then, we requested parents’ and guardians’ consent for the students to participate in this study (only 1% did not give consent). We explained the relevance and the goals of the study to the students and informed them that participation was voluntary and anonymous. The administration of the instruments was carried out by a group of trained researchers with at least one qualified researcher (with a Ph.D.) present during the administration of the instruments to provide students with the necessary support. Measures were collected in the classrooms during a regular class period without the presence of the teachers. In parallel, the homeroom teachers (the teacher who is in charge of taking care of students’ academic and personalrelated affairs in a specific classroom during an academic course) filled in the PROF-A scale. The research was conducted in compliance with ethical standards required for research with human beings, respecting the basic principles included in the Helsinki Declaration and approved by the Ethics Committee of the Vice-Chancellor for Research of the host university. 2.2. Participants The study was an ex post facto cross-sectional and descriptive study. Participants in the study were 2399 adolescents (50% girls), aged between 11 and 18 years (M= 14.65, SD = 1.78), enrolled in five state schools of Secondary Compulsory Education and High School located in the Valencian Community (Spain). Sample distribution by academic grade, family composition, and parents’ educational level is presented in Table 1. 2.3. Instruments 2.3.1. Academic Competence and Teacher–Student Relationship To evaluate academic competence and teacher–student relationship, we used two of the subscales of the School Adjustment Perceived by the Teacher Scale (Escala de Ajuste Escolar Percibido por el Profesor—PROF-A) developed by Cava et al. [ 52 ]. This scale was designed to assess the teachers’ perception of different factors related to their students’ social adjustment and academic performance by using 14 items. Academic competence contains four items (e.g., “Student’s participation in activities, discussions, debates, etc., proposed in class”) and the teacher–student relationship subscale includes four items (e.g., “The time I spend talking to this student”). Items are rated on a 10-point scale ranging from 1 (very low) to 10 (very high). Adequate psychometric properties were obtained in previous studies performed with samples of secondary students [ 52 , 53 ], with Cronbach’s alpha values ranging from 0.85 to 0.92. In our study, the Cronbach’s alpha obtained for academic competence was 0.94 and 0.85 for the teacher–student relationship. In the confirmatory factorial analysis (CFA), the chi-square values and fit indexes were χ2(103) = 1137.31, p< 0.001, RMSEA (root mean square error of approximation) = 0.06 [0.06, 0.07], and CFI (comparative fit index) = 0.96. Int. J. Environ. Res. Public Health 2021,18, 1163 5 of 16 Table 1. Sociodemographic characteristics of the sample. Sociodemographic Characteristics % Academic Grade 1st grade secondary 21.4 2nd grade secondary 20.1 3rd grade secondary 18.2 4th grade secondary 18.6 1st grade high school 10.2 2nd grade high school 11.5 Family Composition Intact 83.5 Single parent 15.7 Living with other adults 0.8 Mother Educational Level Higher education 35.9 High school 16.4 Primary school 25.8 No studies 8 Unknown 13.9 Father Educational Level Higher education 37.8 High school 14.7 Primary school 22 No studies 8.4 Unknown 17.1 2.3.2. Classroom Environment We used the relationship dimension of the classroom environment scale [ 54 ] in the Spanish version of Fernández-Ballesteros and Sierra [ 55 ]. This scale evaluates classroom environment from the point of view of the students through 27 binary-choice (true–false) items, forming three subscales: (1) Involvement (degree of student attentiveness, interest, and participation in class activities; 9 items, e.g., “Students put a lot of energy into what they do here”); (2) Affiliation (degree of friendship and support among students; 9 items, e.g., “Students in this class get to know each other really well”); and (3) Teacher support (amount of help, trust, and friendship the teacher offers to students; 9 items, e.g., “The teacher takes a personal interest in the students”). The Cronbach’s alpha values obtained in previous studies with Spanish adolescent samples ranged between 0.77 and 0.89 (e.g., [41,56]). In our sample, Cronbach’s alpha of the global scale was 0.77 and 0.66, and 0.64 and 0.63 for the three subscales, respectively. In the CFA analysis, the chisquare values and fit indexes were χ2(368) = 1075.66, p< 0.001, RMSEA = 0.03 [0.03, 0.04], and CFI = 0.91. 2.3.3. School Violence This scale by Little et al. [ 57 ] (bidirectional translation into Spanish, using the parallel back-translation procedure of Brislin [ 58 ]) includes 25 items that assess participation in aggressive behaviour towards peers at school over the last 12 months. It is rated on a 4-point Likert-type scale ranging from 1 (never) to 4 (always). The scale evaluates six dimensions of violence within two types of aggressive behaviour: overt or direct and relational or indirect; and three functions of violence: pure, reactive, and instrumental. These dimensions are pure overt (e.g., “I’m the kind of person who hits, kicks, or punches others”); reactive overt (e.g., “If others make me upset or hurt me, I often put them down”); instrumental overt (e.g., “I often threaten others to get what I want”); pure relational (e.g., “I’m the kind of person who says mean things about others”); reactive relational (e.g., “If others have threatened me, I often say mean things about them”); and instrumental relational (e.g., “To get what I want, I often ignore or stop talking to others”). In the present study, Int. J. Environ. Res. Public Health 2021,18, 1163 6 of 16 all items were summed to generate six composite variables that fed the school violence construct in the measurement model. The Cronbach’s alpha values obtained in previous studies with Spanish adolescent samples ranged between 0.64 and 0.87 [ 12 , 56 , 59 ]. In our sample, Cronbach’s alpha of the global scale using these composites was 0.85 (the reliability range for the six subscales varied from 0.63 in instrumental relational to 0.77 in reactive overt). In the CFA analysis, the chi-square values and fit indexes were χ2 (178) = 641.02, p< 0.001, RMSEA = 0.03 [0.03, 0.04], and CFI = 0.91. 2.3.4. School Victimisation The peer–victimisation scale developed by Cava et al. [ 60 ] is a measure of self-reported victimisation based on the multidimensional peer–victimisation scale of Mynard and Joseph [ 61 ] and the social experience questionnaire self-report of Crick and Grotpeter [ 25 ]. This scale consists of 20 items that assess the frequency with which the students had experienced 20 victimizing experiences in the last school year on a response range of 1 (never) to 4 (always). The scale presents a three-factor structure: relational victimisation (e.g., “a peer got angry with me and separated me from my group of friends to prevent me from playing or participating in any activity”); overt physical victimisation (e.g., “a peer hit me to really harm me”); and overt verbal victimisation (e.g., “a peer insulted me”). The psychometric properties of the scale are adequate in previous studies [ 60 , 62 ] and in the present study, with internal consistencies (Cronbach α ) of the three subscales ranging from 0.68 to 0.92 (0.92 in the present study for the total scale and 0.89, 0.63, and 0.84 for the three subscales, respectively). In the CFA analysis, the chi-square value and fit indexes were χ2 (159) = 664.28, p< 0.001, RMSEA = 0.04 [0.03, 0.04], and CFI = 0.93. 2.4. Statistical Analysis Firstly, descriptive statistics (means and standard deviations) and Pearson’s correlations were computed using SPSS (Version 25.0, IBM, Armonk, NY, USA). Secondly, multivariate inferential analyses were conducted using structural equation modelling (SEM) through Mplus 8.4 software [ 63 ]. Complementary analyses were also performed to determine the significance and magnitude of the potential mediating effect [ 64 , 65 ]. Finally, to further check the robustness of the proposed model, we tested it using structural invariance across sex through multigroup analyses. The maximum likelihood estimate with robust standard errors (MLR) was used to correct the non-normality of the data [ 63 ]. The fixed models were evaluated using a combination of indices, including the comparative fit index set and the Tucker–Lewis index (CFI/TLI > 0.90 for a reasonable fit and CFI/TLI > 0.95 for a good fit), the mean square root of the residuals (SRMR < 0.09 for a good fit), and the mean square root of the approximation (RMSEA < 0.06 for a proper fit) [ 66 ]. Although a chi-square test (S-B χ2 ) for model fit was also reported, it was not used to evaluate model fit due to its sensitivity to large sample sizes. Throughout the whole study, the reported path coefficients were standardised values. To perform the test of the indirect effect, its confidence intervals were calculated using the bootstrap method with 2000 samples. 3. Results 3.1. Descriptive Analysis Regarding the descriptive statistics and correlations for the observed variables, Table 2 shows the means, standard deviations, and bivariate correlations. In general, variables of school adjustment perceived by the teacher (academic competence and teacher–student relationship) were positively and significantly related to variables of classroom climate (involvement, affiliation, and teacher support perceived by the students), with rranging from 0.063 to 0.116, p< 0.01. Academic competence and variables of school violence and victimisation were negatively and significantly related (rranging from − 0.040, p< 0.05, for instrumental relational violence to − 0.239, p< 0.01, for reactive overt violence) except for pure relational violence (this bivariate correlation was positive with r= 0.045, p< 0.05). Int. J. Environ. Res. Public Health 2021,18, 1163 7 of 16 Teacher–student relationship was negatively and significantly related to overt reactive violence (r= − 0.085, p< 0.01) and overt physical victimisation (r= − 0.047, p< 0.05), but this relationship was positive with pure violence (both overt and relational; r= 0.043, p< 0.05 and r= 0.077, p< 0.01, respectively). All bivariate correlations between classroom climate and violence and victimisation variables were negative and significant (rranging from −0.051, p< 0.05, to −0.232, p< 0.01). 3.2. Structural Equation Modelling To analyse the direct relationship between school adjustment and school violence and victimisation, and indirectly through classroom climate, a structural equation model (SEM) was created. Table 3shows the latent variables included in the models, their respective indicators, the standard error, and the associated probability for each indicator in the corresponding latent variable. The calculated model (see Figure 1) showed an adequate fit to data: S-B χ2 = 461.304, df = 45, p< 0.001, CFI = 0.952, RMSEA = 0.062 [0.057, 0.067]. A direct and negative relationship was observed between teacher perception of academic competence and school violence ( β = − 0.157, p< 0.001), and between teacher perception of academic competence and school victimisation ( β = − 0.114, p< 0.001). The direct relationship was positive, although smaller, between teacher perception of teacher–student relationship and school violence ( β = 0.084, p< 0.01) and victimisation ( β = 0.052, p< 0.05). Moreover, the three classroom climate variables (involvement, affiliation, and teacher support) were directly and negatively related to school violence (β=−0.069, p< 0.01; β=−0.070, p< 0.01; β = − 0.094, p< 0.001), and two of them (involvement and affiliation) to school victimisation ( β = − 0.129 and β = − 0.199, p< 0.001). The percentage of variability of school violence and victimisation explained by the structural model was 15.1% (9.6% for victimisation and 5.5% for violence), which can be considered to be a size of the effect of the statistical significance of the estimated model. Int. J. Environ. Res. Public Health 2021, 18, x 7 of 17 from 0.063 to 0.116, p < 0.01. Academic competence and variables of school violence and victimisation were negatively and significantly related (r ranging from –0.040, p < 0.05, for instrumental relational violence to −0.239, p < 0.01, for reactive overt violence) except for pure relational violence (this bivariate correlation was positive with r = 0.045, p < 0.05). Teacher–student relationship was negatively and significantly related to overt reactive violence (r = −0.085, p < 0.01) and overt physical victimisation (r = −0.047, p < 0.05), but this relationship was positive with pure violence (both overt and relational; r = 0.043, p < 0.05 and r = 0.077, p < 0.01, respectively). All bivariate correlations between classroom climate and violence and victimisation variables were negative and significant (r ranging from – 0.051, p < 0.05, to −0.232, p < 0.01). 3.2. Structural Equation Modelling To analyse the direct relationship between school adjustment and school violence and victimisation, and indirectly through classroom climate, a structural equation model (SEM) was created. Table 3 shows the latent variables included in the models, their respective indicators, the standard error, and the associated probability for each indicator in the corresponding latent variable. The calculated model (see Figure 1) showed an adequate fit to data: S-B χ 2 = 461.304, df = 45, p < 0.001, CFI = 0.952, RMSEA = 0.062 [0.057, 0.067]. A direct and negative relationship was observed between teacher perception of academic competence and school violence (β = −0.157, p < 0.001), and between teacher perception of academic competence and school victimisation (β = −0.114, p < 0.001). The direct relationship was positive, although smaller, between teacher perception of teacher–student relationship and school violence (β = 0.084, p < 0.01) and victimisation (β = 0.052, p < 0.05). Moreover, the three classroom climate variables (involvement, affiliation, and teacher support) were directly and negatively related to school violence (β = −0.069, p < 0.01; β = −0.070, p < 0.01; β = −0.094, p < 0.001), and two of them (involvement and affiliation) to school victimisation (β = −0.129 and β = −0.199, p < 0.001). The percentage of variability of school violence and victimisation explained by the structural model was 15.1% (9.6% for victimisation and 5.5% for violence), which can be considered to be a size of the effect of the statistical significance of the estimated model. Figure 1. Final structural model with path coefficients and statistical significance. *** p < 0.001; ** p < 0.01; * p < 0.05. Figure 1. Final structural model with path coefficients and statistical significance. *** p< 0.001; ** p< 0.01; * p< 0.05. Int. J. Environ. Res. Public Health 2021,18, 1163 8 of 16 Table 2. Pearson’s correlations and descriptive statistics for observed variables. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1. Academic competence - 2. Teacher–student relationship 0.468 ** - 3. Involvement 0.028 0.116 ** - 4. Affiliation 0.079 ** 0.086 ** 0.310 ** - 5. Teacher support 0.078 ** 0.063 ** 0.334 ** 0.286 ** - 6. Pure overt violence −0.067 ** 0.043 * −0.097 ** −0.105 ** −0.149 ** - 7. Reactive overt violence −0.239 ** −0.085 ** −0.125 ** −0.102 ** −0.118 ** 0.460 ** - 8. Instrumental overt violence −0.090 ** 0.012 −0.067 ** −0.119 ** −0.127 ** 0.504 ** 0.398 ** - 9. Pure relational violence 0.045 * 0.077 ** −0.051 * −0.086 ** −0.119 ** 0.417 ** 0.205 ** 0.412 ** - 10. Reactive relational violence 0.021 0.030 −0.110 ** −0.083 ** −0.119 ** 0.288 ** 0.309 ** 0.302 ** 0.438 ** - 11. Instrumental relational violence −0.040 * 0.004 −0.072 ** −0.106 ** −0.097 ** 0.327 ** 0.257 ** 0.603 ** 0.456 ** 0.401 ** - 12. Relational victimisation −0.066** −0.003 −0.173 ** −0.232 ** −0.125 ** 0.139 ** 0.039 0.160 ** 0.176 ** 0.164 ** 0.186 ** - 13. Overt physical victimisation −0.123** −0.047 * −0.108 ** −0.160 ** −0.093 ** 0.144 ** 0.146 ** 0.139 ** 0.091 ** 0.017 0.092 ** 0.496 ** - 14. Overt verbal victimisation −0.086** −0.039 −0.197 ** −0.222 ** −0.148 ** 0.203 ** 0.117 ** 0.174 ** 0.155 ** 0.117 ** 0.148 ** 0.752 ** 0.604 ** - Mean 5.927 7.388 1.455 1.717 1.597 1.366 1.578 1.133 1.332 1.778 1.192 1.471 1.180 1.593 Standard deviation 2.028 1.313 0.208 0.173 0.220 0.322 0.542 0.243 0.323 0.470 0.305 0.439 0.284 0.470 Note: ** p< 0.01; * p< 0.05. Int. J. Environ. Res. Public Health 2021,18, 1163 9 of 16 Table 3. Factorial loadings, standard error, and associated probability. Factors and Variables Factor Loadings (Standard Errors) School Violence Pure overt violence 1 a Reactive overt violence 1.834 *** (0.432) Instrumental overt violence 0.656 *** (0.073) Pure relational violence 0.750 *** (0.081) Reactive relational violence 0.861 *** (0.101) Instrumental relational violence 0.811 *** (0.087) School Victimisation Relational victimisation 1 a Overt physical victimisation 0.883 *** (0.068) Overt verbal victimisation 0.958 *** (0.068) Note: Robust statistics. Standard errors in brackets. aFixed at 1 during estimation. *** p< 0.001. Regarding the indirect relationships or mediational effects, we calculated all of them and the results showed multiple significant indirect effects of academic competence and teacher–student relationship on school violence and victimisation through the three variables of classroom climate. Their magnitude ranged from − 0.001 (e.g., Academic competence → Involvement → Teacher support → School violence) to − 0.014 (Teacher–student relationship → Involvement → School victimisation); that is, all of them were very small. Total indirect effects, direct effects, and total effects are presented in Table 4. Overall, the total effect of teacher perception of academic competence on school violence and victimization was negative (i.e., protective, higher academic competence was associated to lower violence and victimization) and higher than total effect of teacher perception of teacher– student relationship (in this case total effects were positive, i.e., higher teacher–student relationship was associated to higher violence and victimization). Concerning the effect of teacher perception of academic competence, both direct and indirect effects were negative and, in consequence, the total effects were higher and significant (β=−0.164 for violence and β = − 0.122 for victimization, p< 0.001). Regarding the effect of teacher perception of teacher–student relationship both, direct and indirect effects, were smaller and opposite (a risk if direct and a protection if indirect) and, in consequence, total effects were lower and only significant, as a risk, for school violence (β= 0.067, p< 0.05). Finally, to check the robustness of this model, we tested its structural invariance across sex groups (boys and girls) through a multigroup analysis. Two models were tested: in the unrestricted model, parameter estimates (factor loadings and structural paths) were freely estimated across groups; in the restricted model, we constrained each of the factor loadings and the structural paths to be invariant across groups. If the chi-square of the restricted model was significantly larger than the chi-square of the unrestricted model, the assumption of invariance would not be tenable. The restricted model showed a fit of S-B χ2 = 939.430, df = 139, p< 0.001, CFI = 0.907, RMSEA = 0.69 [0.065, 0.074]. The results showed significant differences between boys and girls in the model ( ∆χ2(35, 2399) = 103.432, p< 0.001). Observing the analyses and using index modification, three paths presented significant differences between the sex groups: first, the relationship between teacher support and school violence was negative and significant in girls ( β = − 0.148, p< 0.001) but higher in boys ( β = − 0.165, p< 0.001); second, the loading of instrumental overt violence in school violence was positive and significant in girls ( β = − 0.448, p< 0.001) but higher in boys ( β = − 0.585, p< 0.001); and finally, the relationship between affiliation and school Int. J. Environ. Res. Public Health 2021,18, 1163 16 of 16 47. Runions, K.C.; Shaw, T. Teacher–child relationship, child withdrawal and aggression in the development of peer victimization. J. Appl. Dev. Psychol. 2013,34, 319–327. [CrossRef] 48. Juvonen, J. Sense of belonging, social bonds, and school functioning. In Handbook of Educational Psychology; Alexander, P.A., Winne, P.H., Eds.; Lawrence Erlbaum Associates Publishers: Mahwah, NJ, USA, 2006; pp. 655–674. 49. Valdés-Cuervo, Á.A.; Martínez-Ferrer, B.; Carlos-Martínez, E.A. The role of teaching practices in the prevention of school violence among peers. J. Psychodidactics 2018,23, 33–38. [CrossRef] 50. Harris, M.J.; Rosenthal, R. Mediation of interpersonal expectancy effects: 31 meta-analyses. Psychol. Bull. 1985 ,97, 363–386. [CrossRef] 51. Rubie-Davies, C. Expecting the Best: Instructional Practices, Teacher Beliefs and Student Outcomes. Ph.D. Thesis, University of Auckland, Auckland, New Zealand, 2004. 52. Cava, M.J.; Povedano, A.; Buelga, S.; Musitu, G. Psychometric analysis of the Scale of Teacher’s Perception of School Adjustment (PROF-A). Interv. Psicosoc. 2015,24, 63–69. [CrossRef] 53. Cañas, E.; Estévez, E.; León-Moreno, C.; Musitu, G. Loneliness, family communication, and school adjustment in a sample of cybervictimized adolescents. Int. J. Environ. Res. Public Health 2020,17, 335. [CrossRef] 54. Moos, R.H.; Trickett, E.J. Classroom Environment Scale Manual; Consulting Psychologist Press: Palo Alto, CA, USA, 1973. 55. Fernández-Ballesteros, R.; Sierra, B. Escalas de Clima Social FES, WES, CIES y CES; TEA: Madrid, Spain, 1989. 56. Jiménez, T.I.; Estévez, E. School aggression in adolescence: Examining the role of individual, family and school variables. Int. J. Clin. Health Psychol. 2017,17, 251–260. [CrossRef] 57. Little, T.D.; Henrich, C.C.; Jones, S.M.; Hawley, P.H. Disentangling the “whys” from the “whats” of aggressive behaviour. Int. J. Behav. Dev. 2003,27, 122–131. [CrossRef] 58. Brislin, R.W. The wording and translation of research instrument. In Field Methods in Cross-Cultural Research; Lonner, W.J., Berry, J.W., Eds.; Sage: Newbury Park, CA, USA, 1986; pp. 137–164. 59. Moreno-Ruiz, D.; Estévez, E.; Jiménez, T.I.; Murgui, S. Parenting style and reactive and proactive adolescent violence: Evidence from Spain. Int. J. Environ. Res. Public Health 2018,15, 2634. [CrossRef] 60. Cava, M.J.; Musitu, G.; Murgui, S. Individual and social risk factors related to overt victimization in a sample of Spanish adolescents. Psychol. Rep. 2007,101, 275–290. [CrossRef] 61. Mynard, H.; Joseph, S. Development of the Multidimensional Peer-Victimization Scale. Aggress. Behav. 2000 ,26, 169–178. [CrossRef] 62. Estévez, E.; Estévez, J.F.; Segura, L.; Suárez, C. The influence of bullying and cyberbullying in the psychological adjustment of victims and aggressors in adolescence. Int. J. Environ. Res. Public Health 2019,16, 2080. [CrossRef] [PubMed] 63. Muthén, L.K.; Muthén, B.O. Mplus User’s Guide, 8th ed.; Muthén & Muthén: Los Angeles, CA, USA, 2017. 64. Mackinnon, D.P.; Dwyer, J.H. Estimating mediated effects in prevention studies. Eval. Rev. 1993,17, 144–158. [CrossRef] 65. Mackinnon, D.; Fairchild, A.; Fritz, M. Mediation Analysis. Annu. Rev. Psychol. 2007,58, 593–614. [CrossRef] [PubMed] 66. Kline, R.B. Principles and Practice of Structural Equation Modeling; The Guilford Press: New York, NY, USA, 2011. 67. Rubie-Davies, C.M.; Peterson, E.R.; Sibley, C.G.; Rosenthal, R. A teacher expectation intervention: Modelling the practices of high expectation teachers. Contemp. Educ. Psychol. 2015,40, 72–85. [CrossRef] 68. Wolke, D.; Woods, S.; Stanford, K.; Schulz, H. Bullying and victimization of primary school children in England and Germany: Prevalence and school factors. Br. J. Psychol. 2001,92, 673–696. [CrossRef] 69. McMillan, J.H.; Myran, S.; Workman, D. Elementary teachers’ classroom assessment and grading practices. J. Educ. Res. 2002 ,95, 203–213. [CrossRef] 70. Peterson, E.R.; Rubie-Davies, C.; Osborne, D.; Sibley, C. Teachers’ explicit expectations and implicit prejudiced attitudes to educational achievement: Relations with student achievement and the ethnic achievement gap. Learn. Instruct. 2016 ,42, 123–140. [CrossRef] 71. Rubie-Davies, C.M.; Rosenthal, R. Intervening in teachers’ expectations: A random effects meta-analytic approach to examining the effectiveness of an intervention. Learn. Individ. Differ. 2016,50, 83–92. [CrossRef] 72. Alberto, P.; Troutman, A.C. Applied Behavior Analysis for Teachers, 9th ed.; Pearson Education: Upper Saddle River, NJ, USA, 2012. 73. Cañas, E.; Estévez, E.; Martinez-Monteagudo, M.C.; Delgado, B. Emotional adjustment in victims and perpetrators of cyberbullying and traditional bullying. Soc. Psychol. Educ. 2020,20, E76. [CrossRef] 74. Troop-Gordon, W.; Kopp, J. Teacher–child relationship quality and children’s peer victimization and aggressive behavior in late childhood. Soc. Dev. 2011,20, 536–561. [CrossRef] 75. Crick, N.R.; Nelson, D.A. Relational and physical victimization within friendships: Nobody told me there’d be friends like these. J. Abnorm. Child Psychol. 2002,30, 599–607. [CrossRef] [PubMed] 76. Bouman, T.; van der Meulen, M.; Goossens, F.A.; Olthof, T.; Vermande, M.M.; Aleva, E.A. Peer and self-reports of victimization and bullying: Their differential association with internalizing problems and social adjustment. J. Sch. Psychol. 2012 ,50, 759–774. [CrossRef] [PubMed] 77. Maxwell, S.E.; Cole, D.A.; Mitchell, M.A. Bias in cross-sectional analyses of longitudinal mediation: Partial and complete mediation under an autoregressive model. Multivar. Behav. Res. 2011,46, 816–841. [CrossRef] [PubMed]