The Influence of Social-Emotional Skills on the Perception of School Violence Among Chilean Children and Adolescents

1. Introduction

The return to in-person schooling following the pandemic marked a turning point in school violence levels in Chile, which have shown a sustained increase ever since, through the end of 2025 (MINEDUC, 2021; SUPEREDUC, 2025). However, the mechanisms through which students experience and interpret this violence in different ways remain poorly understood. In this regard, social-emotional skills (SES) have emerged as a central analytical focus in contemporary educational research, as they provide insight into the processes through which children and adolescents interpret, regulate, and respond to the demands of the school environment. Moving beyond dispositional conceptions, current literature defines them as integrated systems of cognitive-affective and relational processes that organise social experiences in specific contexts (Arango et al., 2024; Denham et al., 2015; Eisenberg et al., 2010). This approach emphasises that SES do not function as static attributes but rather as functional resources whose activation depends on contextual characteristics and the specific demands of social interaction, a point that is particularly relevant in school settings characterised by high relational complexity.

From this perspective, SES play a key role in social interpretation processes. Models of social information processing (Crick & Dodge, 1994; Lemerise & Arsenio, 2000) have shown that behaviour in interpersonal contexts is mediated by a sequence of cognitive and emotional operations, including: the encoding of social cues, the attribution of intentionality, and the evaluation of possible responses. In this context, SES directly influence the way individuals interpret ambiguous situations, regulate their emotional responses, and select strategies for action, thereby shaping the subjective experience of school interactions (Compas et al., 2017; Eisenberg et al., 2010; Herrera-Pulgarín, 2025). This emphasis on interpretive processes allows research to shift the focus from observable behaviour to the mechanisms that organise social experience.

This interpretive dimension takes on particular significance in the study of school violence. Traditionally approached from a perspective centred on direct physical aggression, school violence has gradually been reconceptualised as a relational, contextual, and multidimensional phenomenon that includes verbal, relational, symbolic, and digital forms (Espelage & Swearer, 2010; Hong & Espelage, 2012; Menesini & Salmivalli, 2017). In this context, a significant step forward has been the recognition of school violence’s perceptual nature: the experience of violence does not depend solely on its objective occurrence, but rather on the processes through which students interpret social interactions (Thornberg, 2015; Thornberg et al., 2017). Thus, two students exposed to similar situations may report significantly different levels of perceived violence, making subjective variability a central component of the phenomenon. This perceptual variability is systematically associated with differences in socio-emotional resources.

In fact, higher levels of emotional regulation, adaptive coping, and perceived social support are associated with lower levels of perceived victimisation, even under comparable conditions of exposure (Wang et al., 2026). These findings suggest that SES act as a mediating mechanism that modulates the interpretation of potentially threatening situations, reducing the attribution of hostile intent and mitigating the psychosocial impact of violence. However, the evidence also indicates that these relationships do not follow strictly linear patterns, which poses challenges for traditional explanatory models (Cantor et al., 2019).

In this regard, a persistent limitation in the literature is the prevalence of variable-centred approaches, which assume homogeneous relationships between SES and psychosocial outcomes. However, recent research has challenged this premise, showing that socio-emotional development is organised into heterogeneous configurations or profiles that combine different skills in a non-uniform manner, leading to distinct trajectories of adjustment, vulnerability, and resilience (Cantor et al., 2019; Masten et al., 2021; Mulvey et al., 2022; Zych et al., 2021). This shift toward person-centred approaches provides a more accurate understanding of the complexity of social-emotional development and its impact on the school experience, particularly in phenomena that depend on interpretive processes, such as the perception of violence.

Furthermore, variables traditionally considered explanatory, such as gender, have been proven to have limited scope in accounting for these dynamics. While differences have been documented in certain socio-emotional dimensions, the evidence points to high intra-group variability and a significant overlap of profiles, which calls into question their ability to directly explain the perception of complex phenomena (Hyde, 2014; Zell et al., 2015). Consequently, a need to move toward models that prioritise socio-emotional contexts over isolated demographic categories has emerged.

These considerations are particularly relevant for students in the second cycle of elementary education, a stage characterised by significant changes in emotional regulation, social cognition, and the central role of peer groups (Blair & Raver, 2015; Dahl et al., 2018; Verdugo, 2026). During this period, heightened interpersonal sensitivity and the complexity of relational dynamics increase the importance of interpretive processes, such that differences in socio-emotional resources can amplify or mitigate perceptions of threat, exclusion, or injustice in the school environment.

Despite these advances, a gap remains between models that analyse the direct effects of SES on perceptions of school violence and those that address the heterogeneity of socio-emotional profiles. This disconnection limits understanding of the mechanisms by which SES shapes distinct school experiences, thereby reducing the explanatory power of current models.

In this context, the present study aims to examine the relationship between social-emotional skills and perceptions of school violence among middle school students, while simultaneously considering (a) the direct effects of social-emotional competencies and (b) the heterogeneity in social-emotional development profiles. This approach seeks to contribute to a more integrated understanding of the phenomenon by articulating complementary analytical levels and moving toward models that recognise the complexity of the processes shaping the experience of violence in school settings.

1.1. Social-Emotional Skills and Development in School Settings

SES are a key multidimensional construct for understanding how children and adolescents interpret, regulate, and respond to the demands of the school environment. In contrast to reductionist approaches focused on isolated traits, contemporary research conceptualises them as integrated systems of cognitive-affective and relational processes that enable individuals to recognise and regulate their own emotions, understand the experiences of others, and act adaptively in complex social contexts (CASEL, 2015; Jones et al., 2011; OECD, 2024). From this perspective, SES do not function as static attributes, but rather as situated functional resources whose activation depends on contextual characteristics and the specific demands of social interaction (Avivar-Cáceres et al., 2022; Denham et al., 2015; Eisenberg et al., 2010).

This situated nature is essential to understanding their role in the school experience, given that SES mediate social interpretation processes such as the attribution of intentionality, the assessment of threat, and the regulation of emotional responses to ambiguous or potentially conflictive situations (Crick & Dodge, 1994; Fuentes-Vilugrón et al., 2025; Lemerise & Arsenio, 2000). In this regard, SES constitute social processing systems that organise students’ subjective experiences within contexts of everyday interaction (Denham et al., 2015; Eisenberg et al., 2010; Lemerise & Arsenio, 2000; Muñoz-Troncoso & Riquelme-Mella, 2025).

Recent literature has also emphasised the need to distinguish, without empirically separating them, between skills, competencies, and socioemotional dispositions. While skills refer to specific abilities that can be learned, such as emotional regulation or conflict resolution, competencies involve their functional integration into real-world contexts, while dispositions refer to relatively stable tendencies that influence their activation and use (Duckworth & Yeager, 2015; Kankaraš & Suárez-Álvarez, 2019). These dimensions form an interdependent system in which skills emerge based on individual dispositions and are consolidated into competencies through social experience, thereby reinforcing a non-reductionist view of social-emotional development.

This system takes on particular importance during middle childhood and early adolescence, a period characterised by significant changes in emotional regulation, social cognition, and interpersonal sensitivity (Blair & Raver, 2015; Dahl et al., 2018). At this stage, the strengthening of executive functions leads to greater self-regulation, while the growing centrality of peer groups increases the complexity of relational dynamics and the demands for psychosocial adjustment (Somerville, 2013). However, recent evidence challenges homogeneous evolutionary models, showing that the development of SES is characterised by multiple, asynchronous trajectories, with high levels of intra- and inter-individual heterogeneity (Howard & Ferrari, 2022; Masten & Cicchetti, 2016).

This heterogeneity stems from the interaction of individual, relational, and contextual factors, including school experiences, the quality of relationships with peers and adults, and varying levels of exposure to psychosocial stressors (Cantor et al., 2019; Jones et al., 2019). Thus, students with similar levels of exposure to adverse situations may differ significantly in how they interpret those experiences, a distinction that is particularly relevant for understanding complex relational phenomena such as school violence.

In this context, while the literature has documented gender differences in certain dimensions of social-emotional learning, such as higher levels of empathy in girls, recent studies suggest that these differences have limited explanatory power, given the high degree of intragroup variability and the overlap of social-emotional profiles (Hyde, 2014; Zell et al., 2015). Consequently, it has been proposed to shift toward approaches that focus on socio-emotional profiles and patterns rather than on isolated demographic categories, thus enabling a more accurate depiction of the diversity of developmental trajectories and their impact on the school experience.

From a functional perspective, SES can be understood as psychosocial mediation systems that bridge the relationship between the school environment and students’ subjective experiences (Cantor et al., 2019; Compas et al., 2001; Osher et al., 2020). Factors such as emotional regulation, active coping, perceived social support, and positive expectations interact in an integrated manner, shaping distinct patterns of adjustment and adaptation (Compas et al., 2017; Domitrovich et al., 2017; Osher et al., 2020). This approach allows us to view SES not only as predictors of behaviour, but also as mechanisms that shape the interpretation of social interactions, which is particularly relevant for the analysis of school violence.

1.2. School Violence as a Relational Phenomenon

School violence has been traditionally approached from a behavioural perspective focused on direct physical aggression; however, in recent decades, there has been a conceptual shift toward ecological approaches that view it as a relational, contextual, and multidimensional phenomenon (Espelage & Swearer, 2010; Hong & Espelage, 2012). Olweus’s pioneering work (1993) established fundamental criteria such as repetition and power asymmetry, but contemporary research has significantly expanded this framework by incorporating symbolic, relational, and digital dimensions of violence.

Today, school violence encompasses not only physical acts but also verbal, relational, and structural forms, such as social exclusion, humiliation, symbolic violence, and cyberbullying (Menesini & Salmivalli, 2017; Muñoz-Troncoso et al., 2022; Thornberg, 2015; Utley et al., 2022). These forms of violence operate within complex systems of interaction, where social norms, group hierarchies, and institutional dynamics shape students’ day-to-day experiences.

In this context, it is essential to recognise that school violence cannot be understood solely in terms of its objective occurrence, but necessarily involves its perceptual dimension (Hong & Espelage, 2012; Meldrum et al., 2022; Thornberg, 2015; Thornberg et al., 2017). The experience of violence depends on processes of social interpretation, including the attribution of intent, the assessment of threat, and the interpretation of implicit norms of interaction (Thornberg, 2015). In this regard, two students exposed to the same situation may experience significantly different levels of perceived violence, depending on their socio-emotional and contextual resources.

Thus, different forms of violence –verbal, physical, relational, digital, and perpetrated by adults– have distinct yet converging effects on emotional well-being, perceptions of safety, and a sense of belonging at school (Gaffney et al., 2021; Zych et al., 2021). Furthermore, these forms often coexist and interact, creating cumulative experiences of victimisation that intensify psychosocial impact. From this perspective, school violence is understood as a situated relational phenomenon, the understanding of which requires integrating individual, social, and institutional dimensions. This approach opens the door to analysing not only the risk factors associated with violence but also the protective mechanisms that modulate its perception and impact, among which SES plays a central role.

In light of the above, it can be argued that SES provide a robust explanatory framework for understanding individual differences in the perception of school violence, insofar as these systems act as mediators that influence the interpretation, evaluation, and response to social interactions. Recent evidence indicates that students with higher levels of emotional regulation, adaptive coping, and perceived social support tend to interpret ambiguous situations as less threatening, reporting lower levels of perceived victimisation even in objectively adverse contexts (Espelage & Hong, 2019; Muñoz-Troncoso & Riquelme-Mella, 2025; Wang et al., 2026). These findings suggest that SES not only influence behavioural responses to violence, but also the cognitive-affective processes that shape perceptions of it.

SES are organised into distinct profiles, in which specific combinations of skills are associated with different levels of vulnerability or resilience to school violence (Mulvey et al., 2022; Zych et al., 2021). In this regard, the perception of violence emerges from the dynamic interaction between contextual conditions and individual socio-emotional configurations, which helps explain why students with similar levels of exposure report divergent subjective experiences.

This perspective is particularly relevant for students in the second cycle of basic education, a stage marked by heightened social awareness and increasingly complex peer relationships. In this context, students with uneven socio-emotional development may amplify perceptions of threat, exclusion, or injustice, whereas those with a more well-rounded development tend to employ interpretive and relational strategies that mitigate the psychosocial impact of violence (Gómez-Villalpando & González-García, 2026).

Consequently, analysing the relationship between SEL and perceptions of school violence requires considering not only the direct effects of social-emotional competencies, but also the diversity of profiles and developmental trajectories that shape varied school experiences. Within this framework, the present study aims to examine the relationship between social-emotional skills and perceptions of school violence among students in the second cycle of basic education, considering both the direct effects of social-emotional competencies and the heterogeneity in student profiles. To this end, the following hypotheses are proposed:

H1: Dimensions of social-emotional skills will have negative and statistically significant effects on the perception of school violence.

H2: Gender differences in social-emotional skills will be reflected in the corresponding differences in perceptions of school violence.

H3: Distinct student profiles will be identified based on their social-emotional skills.

H4: At least one profile with higher levels of SES will be identified that reports a lower perception of school violence compared to profiles with lower SES.

2. Materials and Methods

The study employs a quantitative, non-experimental, cross-sectional, and comparative methodology (León & Montero, 2015; McMillan & Schumacher, 2005).

2.1. Participants

A total of 311 students from the school system, aged 8 to 15 years (M=10.65; SD=1.69), participated. Of the total participants, 55.6% were male, and 44.4% were female. At the time of participation, the students were in grades 3 through 8 (Table 1).

2.2. Instruments

A Social-Emotional Skills Questionnaire (SES) was developed for this study, consisting of 16 items that measure four dimensions: [1] stress management (4 items, e.g., “I have been able to calm myself down when something makes me sad or angry”); [2] adaptation (4 items, e.g., “When I’ve had a problem, I try to solve it calmly”); [3] sense of security (4 items, e.g., “I’ve felt that my family protects me”); and [4] expectations (4 items, e.g., “A lot of good things happen to me during the day”). The dimensions are conceptually aligned with different constructs: the stress management dimension relates to models of emotional regulation (Gross & John, 2003); the adaptation dimension relates to active coping strategies, specifically problem-solving and seeking social support (Ayers et al., 1996); the security dimension with scales of perceived family social support (Zimet et al., 1988); and the expectations dimension with constructs of dispositional optimism (Ortiz et al., 2016; Scheier & Carver, 1985). The items are rated on a 6-point Likert scale from 1 (Strongly Disagree) to 6 (Strongly Agree), with higher scores indicating higher levels of SES.

The Abbreviated CENVI Questionnaire (Muñoz-Troncoso et al., 2023), specifically the second-order factor “Types of School Violence” (TSV), a unifactorial model validated by Muñoz-Troncoso and Riquelme-Mella (2025) was also used. This is a scale that measures students’ perceptions of types of school violence, composed of five first-order dimensions: [1] verbal violence (3 items, e.g., “Some students give their classmates irritating nicknames”); [2] physical violence (3 items, e.g., “Some students fight near the school”); [3] exclusion (3 items, e.g., “Some classmates embarrass or ridicule others to damage their image”); [4] cyberbullying (3 items, e.g., “Some students create social media accounts under fake names to spread comments that embarrass others”); and [5] teacher violence (3 items, e.g., “Some teachers grab students firmly by the arms to get their attention”). The scale demonstrated a good fit in the AFC (Muñoz-Troncoso et al., 2023; Muñoz-Troncoso & Riquelme-Mella, 2025), with internal reliability indices ω > .80 across all dimensions, and scalar invariance for gender, ethnicity, administrative affiliation, and grade level (Muñoz et al., 2023). It is a 6-point Likert scale (1=Never – 6=Always) where higher scores indicate a greater perception of TSV.

2.3. Procedure

The study was conducted as part of the project “The Relationship Between Emotional Regulation and the Perception of School Violence Among Students in the Araucanía Region,” which was reviewed and approved by the Research Office of the Catholic University of Temuco (Chile), code 2025PF-FM-01. The school obtained parental consent and ensured students' voluntary participation. The procedure was carried out on the school premises, and the children were accompanied by their teachers during the administration of the questionnaire.

The survey is hosted on a LimeSurvey-based platform (2024), available at www.censa.cl. It begins with a description of the study, an informed consent form, and a privacy notice that includes details about the research, the survey, and the estimated time required to complete it. The voluntary nature of participation was clearly stated, and anonymity and data protection were guaranteed.

The study was conducted in accordance with the international ethical guidelines set forth in the Declaration of Helsinki (World Medical Association, 2013) and the Declaration of Singapore (Resnik & Shamoo, 2011), as well as with the Chilean regulations established in Law 20.120 (MINSAL, 2006).

2.4. Plan of Analysis

The SES scale underwent content validation through expert judgment, with Aiken’s V coefficient (Aiken, 1980) calculated to assess the appropriateness, relevance, and clarity of the items. A value of V ≥ .70 was considered adequate (Penfield & Giacobbi, 2004). For both scales, the normality of all variables (items and mean scores per dimension) was examined in order to select the estimators and hypothesis tests for subsequent analyses. The SES scale underwent confirmatory factor analysis (CFA).

Although recent studies have demonstrated the validity, reliability, and scalar invariance of the CENVI questionnaire across various categories (Muñoz-Troncoso et al., 2022, 2023), it was deemed appropriate to also review the goodness-of-fit of the indices, as was done for the SES scale, using a CFA. This was done as a preliminary step to the analysis of the proposed structural relationship, which was conducted using structural equation modelling (SEM). Thus, for both CFA and SEM, the unweighted least squares mean- and variance-adjusted estimator (ULSMV) was used (Brown, 2015; Muthén & Muthén, 2017). To assess the model's goodness of fit (in CFA and SEM), the following were considered:

(1) the chi-square statistic and its ratio to the degrees of freedom (χ²/df); (2) the Root Mean Square Error of Approximation (RMSEA); (3) the Comparative Fit Index (CFI); and (4) the Tucker-Lewis Index (TLI). Theoretically, a model is considered to have an adequate fit when the χ²/df ratio is equal to or less than 3; the RMSEA is less than .08 –ideally less than .05–; the CFI and TLI indices are equal to or greater than .90 –and preferably .95– (Byrne, 2016; Hu & Bentler, 1999; Kline, 2023). The reliability of each dimension was assessed using McDonald’s omega coefficient (McDonald, 1999), with values ≥ .60 considered acceptable for exploratory research (Hair et al., 2019).

The structural relationship between SES and the perception of TSV was assessed using the gamma coefficient (γ). For interpretation, thresholds of .10 (low), .30 (moderate), and .50 (high) were used (Cohen, 1988). Statistical significance was set at p < .05 (Hair et al., 2019).

In addition, differences between men and women in SES dimensions and perceptions of TSV were explored using Mann-Whitney U tests. This analysis made it possible to examine whether gender constitutes a significant source of heterogeneity in the patterns observed between the two constructs. Statistical significance was set at p < .05.

In addition, a two-stage cluster analysis was conducted for the SES scales to identify latent student profiles based on their socio-emotional competence profiles. This approach allows for an examination of the sample’s heterogeneity beyond the linear relationships between variables (Kaufman & Rousseeuw, 1990). Schwarz’s Bayesian Information Criterion (BIC) was used to determine the optimal number of clusters, and the log-likelihood measure was employed as the distance metric (Chiu et al., 2001). Subsequently, an assessment of whether the identified levels and profiles differed in their perceptions of TSV was conducted using Mann-Whitney U tests for each dimension of violence, with statistical significance set at p < .05.

The analyses were performed using MPlus v.8.8 (Muthén & Muthén, 2024), SPSS v.25 (IBM Corp., 2020) and Excel v16.74 (Microsoft, 2023).

3. Results

Content validation of the SES questionnaire was conducted with seven experts. Aiken’s V coefficients ranged from .78 to 1 for relevance, from .81 to 1 for significance, and from .75 to .98 for clarity. All items met the criterion of V ≥ .70. The Kolmogorov–Smirnov test showed that the data do not follow a normal distribution (p < .001). Regarding the AFCs, a good fit of the proposed model to the data was found for both the SES scale (χ² = 261.415; df = 98; RMSEA = .073; CFI = .944; TLI = .932) and for the TSV scale (χ² = 160.914; df = 85; RMSEA = .054; CFI = .975; TLI = .969).

The composite reliabilities of the SES dimensions were as follows: stress management (ω = .645), adaptation (ω = .710), sense of security (ω = .885), and expectations (ω = .799). For TSV, these results were: verbal violence (ω = .820), physical violence (ω = .822), exclusion (ω = .864), cyberbullying (ω = .883), and teacher violence (ω = .860)

The SEM analysis revealed that the proposed model fit the data well (χ² = 578.592, df = 419, RMSEA = .035, CFI = .961, TLI = .957). Furthermore, a statistically significant moderate effect was observed for each latent SES variable on the second-order latent variable of TSV, as detailed in Table 2.

The comparison by gender showed that men scored significantly higher on stress management (p = .011), safety (p = .044), and expectations (p = .045). The difference in adaptation did not reach statistical significance, although it was marginally above the threshold (p = .055). Table 3 presents the detailed statistics.

Regarding perceptions of TSV, no statistically significant differences were found between men and women in any of the five dimensions assessed (p > .05). However, men reported higher average scores on all TSV measures. Table 4 presents the detailed statistics.

A two-stage cluster analysis using the four dimensions of SES identified three profiles based on Schwarz’s Bayesian information criterion (BIC = 568.027). The quality of the clustering was satisfactory (silhouette coefficient = 0.6). Profile 1 (n=151; 48.6%) was characterised by high scores on all SES dimensions. Profile 2 (n=134; 43.1%) had intermediate scores. Profile 3 (n=26; 8.4%) had the lowest scores on all dimensions. Statistically significant differences were found in the distribution of profiles by gender (χ² = 7.563, df = 2, p = .023), with a higher proportion of men in Profile 1 (63.6% men vs. 36.4% women), while Profile 2 (48.5% men vs. 51.5% women) and Profile 3 (46.2% men vs. 53.8% women) both showed a higher proportion of women. Table 5 presents the detailed centroids for each profile.

In the Mann-Whitney U tests for all dimensions of TSV, the results showed statistically significant differences between Cluster 1 and Cluster 2 (p < .05), with Cluster 2 reporting a higher perception of violence. Comparisons between Cluster 1 and Cluster 3 revealed statistically significant differences in three of the five dimensions: verbal violence (p = .044), exclusion (p = .009), and digital violence (p = .043), while physical violence (p = .076) and violence by teachers (p = .051) showed marginally insignificant differences. Comparisons between Cluster 2 and Cluster 3 did not yield statistically significant differences in any dimension of TSV. Table 6 presents the detailed statistics for these comparisons.

4. Discussions

The aim of this study was to examine the relationship between social-emotional skills (SES) and perceptions of school violence among middle school students. Overall, the results confirm that SES function as psychosocial mediating resources that influence how students interpret relational dynamics in the school environment.

First, the confirmation of H1 provides evidence consistent with the literature that conceptualises SES as systems that influence the cognitive-affective evaluation of social experiences (Ordoñez Rodríguez, 2022; Osher et al., 2020; Panayiotou et al., 2019). All dimensions of SES had negative, statistically significant effects on perceptions of school violence, with moderate magnitudes. This pattern suggests that students with greater resources for emotional regulation, coping, perceived support, and positive expectations tend to interpret school interactions as less threatening, even in potentially adverse contexts.

The moderate magnitude of the effects is consistent with the multidimensional nature of the perception of school violence, which does not depend exclusively on individual variables, but rather on the interaction between personal, relational, and institutional factors (Malti et al., 2018; Taylor et al., 2017). In this sense, SES do not eliminate exposure to violence but rather modulate the interpretive processes through which it is perceived and evaluated.

One relevant finding is that the expectations dimension showed the most robust effect. This finding suggests that dispositional optimism and the anticipation of positive outcomes constitute particularly relevant resources for the reinterpretation of ambiguous events, reducing the attribution of harmful intent and the perception of threat (Compas et al., 2017). This result reinforces the importance of considering anticipatory cognitive dimensions within intervention models for school coexistence.

Regarding H2, the results indicate that, although gender differences exist in some dimensions of SES, these do not translate into differences in the perception of school violence. This finding suggests that gender, as a categorical variable, has limited explanatory power in this type of phenomenon, which is consistent with evidence highlighting high intragroup variability and the overlap of socioemotional profiles (Hyde, 2014; Löffler & Greitemeyer, 2023). Consequently, the relationship between SES and perceptions of violence appears to be better explained by individual configurations rather than by gender differences per se.

In this regard, the confirmation of H3 and H4 provides particularly relevant evidence. The identification of three distinct SES profiles–high, medium, and low–and their association with different levels of perceived violence reinforces the need to move from variable-centred models toward person-centred approaches. The finding that the profile with the highest levels of SES consistently reports lower perceptions of violence is consistent with the literature describing these competencies as protective factors against psychosocial vulnerability (Domitrovich et al., 2017; Zych et al., 2021).

However, the results also show that this relationship is not entirely linear. Differences between profiles are more consistent at high and medium levels of SES than at medium and low levels, suggesting that the protective effect of SES may operate as a threshold rather than in an incremental manner. This pattern is consistent with approaches that propose that certain minimum levels of socioemotional resources are necessary for effective regulation and interpretation strategies to be deployed (Cantor et al., 2019; Osher et al., 2020).

Furthermore, the analysis by type of violence indicates that SES are more consistently associated with perceptions of relational, verbal, and digital forms of violence, while their effect is less clear in cases of physical violence and violence perpetrated by teachers. This finding suggests that SES primarily operate on forms of violence that require greater interpretive processing, where ambiguity and the attribution of intentionality play a central role (Thornberg et al., 2017; Utley et al., 2022). In contrast, more direct or structural forms of violence appear to depend more on contextual conditions and power relations, which limit the effect of individual resources.

It is important to consider certain limitations. In particular, the small sample size of the lowest SES profile limits the statistical power to detect subtle differences between groups, suggesting that the absence of differences between the medium and low profiles should be interpreted with caution. Furthermore, the cross-sectional design of the study precludes causal inference; therefore, future research should incorporate longitudinal designs to examine the directionality of these relationships.

Finally, the results show that SES operate as functional systems of interpretation rather than as static attributes, and that their influence on the perception of school violence is organised in a heterogeneous and context-dependent manner (Carmona-Rojas et al., 2023). This reinforces the need to design interventions that strengthen individual competencies, taking into account the relational and structural dynamics of the school environment.

5. Conclusions

The results confirm that social-emotional skills (SES) function as adaptive systems that mediate the interpretation of school violence, rather than as static individual attributes. Their negative effect on the perception of violence supports their role as cognitive-emotional filters in the school experience.

Furthermore, the findings highlight the need to complement variable-centred approaches with person-centred approaches. The identification of distinct profiles and their nonlinear relationships with perceptions of violence reinforces the importance of considering the heterogeneity of social-emotional development.

Regarding gender, the results suggest an indirect and non-determinative role, calling into question its use as a linear explanatory variable and highlighting its influence on the configuration of social-emotional profiles rather than on the perception of violence itself.

Finally, the protective effect of SES varies depending on the type of violence: it is more consistent in relational and ambiguous forms, where interpretive processes are involved, and less so in direct or structural violence. This has relevant implications for intervention, suggesting that strengthening SES may be particularly effective in mitigating the psychosocial impact of perceived violence, while other forms require structural approaches.

However, these results must be interpreted with certain limitations in mind. The cross-sectional design prevents the establishment of causal relationships between variables, while the small size of some subgroups, particularly those with lower socioemotional development, limits the statistical power to detect finer differences. Furthermore, the use of self-report measures may introduce biases associated with social desirability or subjective interpretation of the items.

Based on the above, future research should incorporate longitudinal designs that allow examination of the directionality of the relationship between SES and perceptions of violence, as well as expand the analysis to different sociocultural and educational contexts. Similarly, it is important to delve deeper into profile-centred approaches and explore the role of contextual variables, such as school climate, teaching practices, and community dynamics, in shaping these relationships. Finally, it is suggested that researchers move toward integrated models that link socio-emotional interventions with institutional strategies to simultaneously address individual interpretive processes and the structural conditions that underpin school violence.