Longitudinal study of peer bullying victimization and its psychological effects on adolescents

Abstract

BACKGROUND

Peer bullying victimization (PBV) is a significant public health issue that adolescents often face, with approximately one-third having experienced PBV. Understanding its interrelationships with mental health problems is crucial for effective intervention. This study aims to examine the longitudinal relationships between internalizing problems (depression and anxiety), externalizing problems (aggression), and PBV among middle school students using cross-lagged panel network analysis.

AIM

To examine the longitudinal relationships between internalizing problems (depression and anxiety), externalizing problems (aggression), and PBV among middle school students using cross-lagged panel network analysis.

METHODS

A total of 1260 middle school students (54.6% male) participated in this study. Data were collected at two time points (time 1 and time 2) using self-report questionnaires to assess PBV, depression, anxiety, and aggression. Cross-lagged panel network analysis was applied to examine the bi-directional relationships between these variables over time.

RESULTS

Depression, particularly a “sad mood,” was the most significant predictor of both PBV and aggression. Cyberbullying victimization also emerged as a key factor influencing depression and anxiety. While anxiety had weaker and less consistent effects on PBV, aggression was predominantly influenced by depressive symptoms and sleep disturbances. The analysis also identified key symptoms, such as a “sad mood” and sleep disturbances, as crucial targets for intervention to mitigate the cycle of PBV, depression, anxiety, and aggression.

CONCLUSION

This study provides important insights for bullying victimization prevention and intervention strategies: (1) Early identification and intervention targeting depression, particularly a “sad mood,” and sleep disturbances; (2) The importance of addressing cyberbullying as a distinct factor influencing mental health; and (3) The need for comprehensive, longitudinal approaches to understanding and intervening in the interconnected mental health issues among adolescents.

Key Words: Peer bullying victimization; Depression; Anxiety; Aggression; Adolescents; Longitudinal study

Core Tip: In the bi-directional relationship between depression and peer bullying victimization (PBV) in adolescents, depression dominates, with sad mood playing a key role in the bi-directional relationship between depression/anxiety and aggression in adolescents, depression/anxiety dominates, with sleep/feeling afraid symptom playing a key role. In the bi-directional relationship between PBV and aggression in adolescents, PBV dominates, with physical bullying victimization playing a dominant role.

INTRODUCTION

Middle school students undergo rapid development across cognitive, emotional, and social domains[1], and this development contributes to heightened social pressures and emotional challenges. During this developmental stage, students face a strong desire for peer acceptance, increased sensitivity to social status, and emotional instability, all of which make them particularly vulnerable to peer bullying[2]. Peer bullying is defined as repeated exposure to aggressive behaviour from peers; such behaviour can manifest physically (e.g., hitting), verbally (e.g., name-calling), socially (e.g., exclusion), or through the spread of harmful rumours, both in person and online[3]. The inclusion of cyberbullying is significant, as it reflects the evolving nature of bullying behaviours in the digital age, where technology has expanded the reach and impact of bullying, making it more pervasive and harder for victims to escape. A meta-analysis revealed that approximately one-third of adolescents experience victimization by peers[4]. In China, the prevalence of peer bullying victimization (PBV) among middle school students ranges from 9.0% to 61.3%[5,6] emphasizing the need for targeted interventions.

A substantial body of research has consistently demonstrated a strong association between PBV and various internalizing problems, particularly anxiety and depression[7]. For example, a study reported that bullied students were more likely to develop depression than their peers were[8]. Moreover, a study identified PBV as a significant predictor of anxiety in adolescents[9]. Longitudinal studies reinforce findings regarding the persistent and damaging effects of PBV on mental health. One study demonstrated that individuals who experienced bullying during childhood were four times more likely to develop severe depression 10 years later[10]. In addition to the effect of PBV on internalizing problems, adolescents with internalizing problems may be at heightened risk of victimization. For example, a study reported that adolescents with depression and anxiety are often perceived as less socially competent and likable, making them more vulnerable to becoming targets of bullying[11]. This finding suggests a reciprocal relationship between internalizing problems and PBV, which is further supported by other reports[12,13], as well as by a meta-analysis[3].

While internalizing problems such as anxiety and depression are frequently associated with PBV, other research also links PBV to externalizing problems[14]. Victims of bullying may express their distress outwardly, engaging in violent behaviours, carrying weapons, or becoming perpetrators of bullying themselves. For example, a study reported that students exposed to PBV were twice as likely to engage in school fights than their peers were[15]. Similarly, a longitudinal study of 2462 Chinese adolescents revealed that prior exposure to PBV significantly predicts subsequent aggression[16], further emphasizing the long-term effect of bullying on externalizing behaviours. Some research also suggests that externalizing problems, such as aggression, may not only be the result of PBV but also contribute to becoming a target for bullying. For example, one study demonstrated that adolescents who exhibit more aggressive behaviours tend to provoke their peers and make them more likely to be targeted by bullying[17]. A longitudinal study of 567 middle school students in Hong Kong reported that PBV and aggression reciprocally served as risk factors for one another. These findings support the idea of a bi-directional relationship between aggression and PBV[18].

Interestingly, the relationship between internalizing and externalizing problems is also bi-directional[19]. For example, certain depressive symptoms, such as irritability, can manifest as aggression and rule-breaking behaviours in social interactions[20]. Conversely, failures in social interactions due to aggression or disruptive behaviours can exacerbate depressive symptoms[21]. To date, we have identified three key bi-directional relationships: PBV with internalizing symptoms (e.g., depression/anxiety), PBV with externalizing problems (e.g., aggression), and internalizing symptoms with aggression. These findings raise several critical questions: How do these interrelationships influence one another? If these relationships are interconnected, it is necessary to control for the variables at play when examining the relationship between any two variables. Furthermore, which aspect of a bi-directional relationship exerts a stronger influence remains unclear. For example, in the relationship between PBV and depression, does prior PBV lead to subsequent depressive symptoms, or does preexisting depression increase the likelihood of PBV?

Given the complexity of these relationships, it is essential to recognize that the concepts involved are multifaceted, encompassing heterogeneous factors. PBV, for example, includes various forms, such as verbal, relational, physical, and cyberbullying, whereas aggression can manifest in physical, verbal, or self-directed behaviours. Similarly, depression and anxiety symptoms differ widely in their presentation. Therefore, when investigating a specific bi-directional relationship, such as that between PBV and depression, it is crucial to determine which type of bullying and which depressive symptoms are most influential in shaping the dynamics at play. These complexities underscore the need for further research to explore the intricate interrelationships among PBV, internalizing problems, and externalizing problems, given that understanding these connections is vital for developing effective intervention strategies.

In recent years, the network analysis approach has emerged as a promising method for understanding the complex relationships among aspects of mental health. Unlike traditional methods, which often treat problems as isolated variables or latent constructs, network analysis makes it possible to examine symptoms and the interactions between symptoms. This approach provides a more nuanced understanding of how specific symptoms influence each other, capturing the complexity of psychological phenomena that traditional methods may overlook. Rather than conceptualizing these problems as distinct, latent entities, network analysis views mental health disorders as arising from causal interactions among a multitude of symptoms[22]. This perspective makes it possible to explore relationships at the level of individual symptoms, offering valuable insights into the interplay and mutual influences among different mental health conditions. A network consists of nodes, which represent observable symptoms or variables, and edges, which represent the associations between nodes[23]. The relationships between symptoms can be quantified and analysed to reveal which symptoms play central roles in the overall network and how they interact with one another.

Unfortunately, to the best of our knowledge, no studies have included PBV, aggression, and depression/anxiety in a network model that examines the interactions among them. To address these shortcomings, cross-lagged panel network (CLPN) analysis might be a good choice[24]. A CLPN is a statistical model within the framework of network theory[25]. CLPN analysis is appropriate for longitudinal design[22], it allows elements to predict with each other, and captures the dynamic relationship between variables by combining autoregression and cross-lagged effects to explore the direction of change between variables over time, on the basis of which we can examine which element is more important in a bi-directional relationship and count the key elements. Therefore, this study constructs a co-occurrence network for PBV, aggression, and depression/anxiety symptoms with the help of CLPN analysis on the basis of two surveys of Chinese middle school students to illustrate the interactions among these variables.

MATERIALS AND METHODS

Participants and procedure

Ethical approval for the current study was granted by (removed for blind review). Written consent was obtained from both parents and adolescents prior to the start of interviews. The surveys were administered at a middle school in Shenzhen, Guangdong Province, China, using an online survey platform. QR codes or links were distributed to the students’ guardians, who were requested to supervise their children to complete the questionnaires independently. The initial survey was conducted from September 5 to September 12, 2022 [time 1 (T1)], followed by a second survey from February 13 to February 20, 2023 [time 2 (T2)], resulting in an approximate six-month interval between the two. Consent for participation was obtained from the local school bureau, the school principal, students, and their parents, all of whom provided signed informed consent forms.

Of the 1416 students who participated in T1, 34 were excluded from the analysis because they did not respond truthfully to each item with due consideration. Among the remaining 1382 students, 85 did not participate in T2, and 37 reported non-honest responses. Therefore, 1260 students provided valid data for both surveys. This group consisted of 676 boys (53.7%), 433 7^th graders (34.4%), and 441 8^th graders (35.0%), with an average age of 13.22 years (SD = 0.95) at T1.

Measures

Depression and anxiety: Depression was assessed using the Chinese version of the 9-item patient health questionnaire (PHQ-9)[26], a self-report measure rated on a scale ranging from 0 (none) to 3 (almost every day). In the present study, Cronbach’s α for the PHQ-9 was 0.89 at T1 and 0.90 at T2. Anxiety was measured using the Chinese version of the 7-item generalized anxiety disorder scale (GAD-7)[27], also rated on a scale ranging from 0 (none) to 3 (almost every day). Cronbach’s α for the GAD-7 was 0.89 at T1 and 0.93 at T2.

Aggression: Aggression was assessed using Buss and Perry’s (1992) aggression questionnaire[28]. The questionnaire contains five factors: Physical aggression, verbal aggression, self-aggression, anger, and hostility. For this study, only the first three factors (physical aggression, verbal aggression, and self-aggression) were included, resulting in a total of 17 items. Each item was rated on a scale ranging from 1 (not compliant) to 5 (fully compliant). Cronbach’s α for this measure was 0.89 at T1 and 0.90 at T2.

PBV: PBV was measured using six items from the child bullying questionnaire[29], revised by[30]. Additionally, four items were used to assess cyberbullying victimization (e.g., “A peer sent me mean or hurtful messages via WeChat, QQ, or mobile text messaging”). The total measure consisted of 10 items, scored on a 5-point scale ranging from 0 (never) to 4 (multiple times in one week). Cronbach’s α was 0.84 at T1 and 0.90 at T2.

Data analysis

Network estimation and visualization: A CLPN model[24] was constructed to illustrate the interaction between variables. The CLPN model was estimated using a series of regressions to calculate autoregressive and cross-lagged coefficients (a variable at T1 predicting another variable at T2 after controlling for all other variables at T1). Ten-fold cross-validation was used to adjust the parameter selection to minimize false-positive edges in the network by regularizing the regression coefficients using least absolute shrinkage and selection operator, and a λ penalty value that minimizes the mean cross-validation error was used. K-fold cross-validation, which is a widely used technique in machine learning, was performed to evaluate the model’s predictive accuracy and generalizability. This method involves dividing the dataset into k equally sized folds, using k-1 folds for training and the remaining folds for testing. The process is repeated k times, with each fold serving as the test set once.

Key elements: There are different subnetworks within the CLPN, and the key element in each subnetwork is of interest in this study. For example, when considering the relationship between depression and PBV, two subnetworks are created, one with edges pointing to PBV from depression and the other with edges pointing to depression from PBV. When exploring the key element of the former, it is necessary to calculate which of the 9 symptoms of depression has the greatest sum of the weights of all the connecting edges between that symptom and PBV, as it is this symptom that is the key element.

Accuracy and stability: We examined the accuracy and stability of the network with two bootstrap methods, as suggested by[22]. First, we estimated the accuracy of the edge weights by performing a nonparametric bootstrap on 1000 instances and calculating the 95% confidence intervals (CI) for each edge. The wider the 95%CI for an edge, the more caution is needed in its interpretation. Second, we used casedrop bootstrapping to estimate correlation stability (CS) coefficients to determine the stability of critical elements. The CS coefficient should be greater than 0.25 and preferably greater than 0.50. Third, whether the key element differences between nodes and the edge weights between edges were significantly nonzero (α = 0.05) was determined.

Key variables in the bi-directional relationship: Still taking the bi-directional relationship between depression and PBV as an example, to determine whether depression has a greater effect on PBV or whether PBV has a greater effect on depression, it is necessary to count the total weight of the edges (W1) emanating from depression and pointing to PBV and the edges (W2) emanating from PBV and pointing to depression. In addition, if W1 is greater than W2, then depression can be considered to play a key role, and vice versa. In addition, with the help of bootstrapping, we performed a Mann-Whitney U test for the difference between W1 and W2 and used Vargha and Delaney’s A to measure the effect size[31], with effect sizes from 0.34-0.63 indicating small effects, effect sizes from 0.64-0.71 or 0.29-0.33 indicating moderate effects, and effect sizes ≥ 0.71 or ≤ 0.29 indicating large effects.

RESULTS

This study used the Harman single-factor method to test for common method bias. A total of 5 common factors greater than 1 appeared, and the first factor accounted for 28.3% of the variance, which is less than the critical value of 40%. Therefore, there was no significant common method bias in this study.

Figure 1 illustrates the co-occurrence network for PBV, aggression, depression, and anxiety among students, providing a visual representation of the CLPN. In this network, solid blue lines represent positive regression coefficients, and arrows indicate unique longitudinal relationships between variables while accounting for all other variables at T1. The autoregressive edges (mean edge weight = 0.19) were notably stronger than the temporal pairwise edges (mean edge weight = 0.04). For improved visual interpretability, autoregressive edges were excluded from Figure 1, as the mapping algorithm scales edge thickness relative to the strongest edge. Table 1 presents the key variables and elements of the network.

Figure 1 The cross-lagged panel network for peer bullying victimization, aggression, depression, and anxiety. Note, the arrow indicates a unique longitudinal relationship from time 1 (September 5 to September 12, 2022) to time 2 (February 13 to February 20, 2023). PBV: Peer bullying victimization; Anx: Anxiety; Agg: Aggression; Dep: Depression.

Table 1 The dominant role and key elements.

Sub-networks	Total weight	|Z|	VDA	Key elements
DEP -> PBV1	0.60 (0.44-0.82)2	31.13c	0.90	DEP 02 (sad mood)
PBV -> DEP	0.07 (0.18-0.40)			PBV 04 (cyberbullying victimization)
ANX -> PBV	0.06 (0.04-0.28)	0.25	0.50	ANX 02 (uncontrollable worrying)
PBV -> ANX	0.02 (0.01-0.25)			PBV 04 (cyberbullying victimization)
DEP -> AGG	1.53 (1.20-1.78)	38.24c	0.99	DEP 03 (sleep)
AGG -> DEP	0.23 (0.18-0.24)			AGG 03 (self-aggression)
ANX -> AGG	0.67 (0.54-0.98)	36.83c	0.98	ANX 07 (feeling afraid)
AGG -> ANX	0.14 (0.11-0.17)			AGG 03 (self-aggression)
AGG -> PBV	0.06 (0.04-0.08)	25.18c	0.18	AGG 02 (verbal aggression)
PBV -> AGG	0.30 (0.13-0.46)			PBV 03 (physical bullying victimization)

^cP < 0.001.

¹The effect of depression (time 1) on peer bullying victimization (time 2).

²0.60 is the total weight of the edges in the cross-lagged panel network emanating from the depression and pointing to the peer bullying victimization, 0.44 is the 25^th percentile of 1000 bootstraps and 0.82 is the 75^th percentile.

DEP: Depression; PBV: Peer bullying victimization; ANX: Anxiety; AGG: Aggression; VDA: Vargha and Delaney’s A.

Table 1 shows that depression played a significantly larger role in its bi-directional relationship with PBV (0.60 vs 0.07, |Z| = 31.13, P < 0.001), while anxiety showed no significant effect in its relationship with PBV (0.06 vs 0.02, |Z| = 0.25, P > 0.05). Both depression (1.53 vs 0.23, |Z| = 38.24, P < 0.001) and anxiety (0.67 vs 0.14, |Z| = 36.83, P < 0.001) showed substantial effects in their bi-directional relationships with aggression. In the relationship between PBV and aggression, PBV had a greater influence (0.30 vs 0.06, |Z| = 25.18, P < 0.001). The effect size analysis indicated that, with the exception of the anxiety-PBV relationship, a key variable could be identified in each of the four remaining bi-directional relationships, with all having large effect sizes.

The key elements differed across the four subnetworks involving depression or anxiety and their relationships with PBV or aggression. However, the key elements remained consistent in the subnetworks extending from PBV or aggression to depression or anxiety: PBV04 (cyberbullying victimization) was the key element for PBV, and self-aggression was identified for aggression. In the subnetworks between PBV and aggression, the key element from aggression to PBV was verbal aggression, whereas PBV03 (physical bullying victimization) was identified as the key element from PBV to aggression.

The CS (stability) coefficient for the network was 0.21, which is below the threshold of 0.25, indicating that caution should be exercised when interpreting the key elements. Appendix Supplementary Figure 1 further supports this caution. Appendix Supplementary Figure 2 shows no significant differences between the critical elements of most nodes, whereas Appendix Supplementary Figure 3 presents the bootstrapped 95%CIs of the edges, suggesting that the accuracy of the co-occurrence network is acceptable. Appendix Supplementary Figure 4 shows the bootstrap difference test results between edges, revealing some significant comparisons.

DISCUSSION

This study innovatively integrates PBV, depression, anxiety, and aggression into a single longitudinal model for middle school students. By employing CLPN analysis, it examines the complex bi-directional relationships and key elements among these variables over time. The findings reveal that depression, especially a “sad mood”, significantly predicts both PBV and aggression, while cyberbullying victimization is a crucial factor influencing depression and anxiety. The study also identifies specific symptoms, such as sleep problems, as key targets for intervention to mitigate the cycle of PBV, depression, anxiety, and aggression. This approach provides a novel perspective on the dynamic interactions between psychological phenomena in adolescents, offering potential directions for effective interventions.

Relationship between PBV and depression/anxiety

Depression exhibited a stronger effect in its bi-directional relationship with PBV. One possible explanation is that depression can impair social competencies, complicating students’ ability to integrate into peer groups and increasing the risk of isolation and being bullied. Depressed students may also suffer from reduced self-esteem, which could undermine their resilience when confronting bullying. Consequently, these students may be more likely to acquiesce and withdraw when faced with bullying[32]. Among the nine depressive symptoms examined, a “sad mood” had the most pronounced effect on PBV. Self-esteem theory holds that sadness, which is a core symptom of depression, can lead to social disengagement and reduced self-esteem, making individuals more vulnerable to feelings of helplessness and despair and, thus, easier targets for bullying[11,33,34].

In contrast, the role of anxiety in the bi-directional relationship with PBV remains less clear. Heightened anxiety may lead some students to avoid peer interactions, potentially reducing their likelihood of becoming victims of bullying. However, our findings suggest that cyberbullying victimization had the most significant influence on depression and anxiety among the forms of PBV. The anonymous nature, potential for widespread dissemination, and persistence of cyberbullying pose unique challenges for intervention and prevention[35]. In middle school contexts, where cyberbullying often targets personal identity, appearance, and personality traits, the impact on victims’ self-esteem can be significant, leading to increased depression and anxiety[4].

Relationship between aggression and depression/anxiety

Depression and anxiety played predominant roles in their interactions with aggression, which is consistent with the findings of[36,37]. This association may be attributed to the impact of depression and anxiety on neurocognitive function[38], particularly in terms of difficulties with emotion regulation and impulse control[39], which may heighten irritability and anger. As a result, students experiencing depression or anxiety may become increasingly frustrated, helpless, or fearful, potentially resorting to aggression as a way to alleviate these emotions. Notably, among the nine depressive symptoms, “sleep disturbances” had the most substantial impact on aggression. Sleep problems can impair emotional regulation, resulting in heightened irritability and aggressive tendencies[40]. Sleep disturbances may also hinder cognitive functioning[41], increasing the likelihood of misinterpreting others’ intentions as hostile. Moreover, in the Chinese educational environment, students often face tremendous academic stress, long hours of study, and high expectations from parents and teachers. This environment can lead to chronic sleep deprivation and poor sleep quality, which in turn exacerbate symptoms of depression and anxiety. Among the seven anxiety symptoms, “feeling afraid” was the most significant predictor of aggression, underscoring the critical role that fear plays in driving aggressive behaviour among students[42]. When students perceive threats, they may use aggression as a self-protective mechanism, masking their underlying fears and insecurities with displays of toughness.

Relationship between PBV and aggression

In the bi-directional relationship between PBV and aggression, PBV demonstrated a stronger effect, with physical bullying playing a particularly pivotal role. Drawing on general strain theory, stress can serve as a precursor to maladaptive coping behaviours[43]. Compared to other forms of PBV, physical bullying may induce greater trauma, anger, and frustration. In response, students may exhibit increased aggression or adopt protective measures to prevent further victimization. This behaviour is also consistent with social interaction theory, which states that there is a vicious cycle of social interaction between the bully and the victim. The passive response of the victim may exacerbate the aggressiveness of the bully, and the aggressiveness of the bully will exacerbate the anxiety and depression of the victim[44]. Additionally, consistent with our study’s findings, a meta-analysis suggested that prior aggression predicts future experiences of PBV, with verbal aggression being particularly influential[45]. In school environments, verbal aggression can spread rapidly, damaging victims’ reputations and social standing. This damage can increase the perpetrator’s risk of broader social isolation and rejection, contributing to an environment that is conducive to peer bullying.

Limitations

Despite the innovative contributions of this study in explaining the relationships between PBV, depression, anxiety, and aggression among middle school students, several limitations must be acknowledged. First, the sample was limited to one geographical region in China, which means that the sample may not be fully representative of broader adolescent populations. The findings may be influenced by cultural and regional factors, and thus, the generalizability of the results to other regions or cultures is uncertain. Future studies should aim to replicate these findings in diverse cultural and geographical settings to enhance the external validity of the conclusions. Second, the study relied solely on self-reported data from students, which introduces the possibility of response bias. Students may underreport or overreport their experiences and feelings due to social desirability, memory lapses, or other factors. To obtain a more comprehensive and accurate picture, future research should incorporate multiple data sources. Such sources could include reports from parents and teachers, who can provide additional insights into students’ behaviours and mental states. Third, the CLPN used in this study had a stability coefficient (CS) of 0.21, which is below the commonly accepted threshold of 0.25. This situation suggests that the key findings should be interpreted with caution, as the stability of the network may not be robust enough to support strong conclusions. Fourth, the sample size of 1260 students is slightly smaller than that of comparable research, which often includes over 1500 participants. This discrepancy may have affected the robustness of our results and the power to detect significant effects. Fifth, the study employed only two survey waves, limiting the use of panel vector autoregression models that could effectively distinguish within-person effects from between-person effects. More frequent assessments over a longer period would provide a richer dataset and allow for a more nuanced understanding of the dynamic relationships between the variables. Finally, while CLPN analysis provides valuable insights into symptom interactions, it cannot establish definitive causal inferences. Future studies could combine experimental or intervention designs to validate the causal relationships between symptoms and enhance clinical intervention effectiveness. In future research, building upon this foundation by incorporating more advanced analytical techniques, such as time-series network analysis, could better capture the dynamic changes across symptoms.

Implications for practice

The findings of this study provide practical insights for designing targeted interventions to address the psychological issues of PBV. By identifying key symptoms, such as a “sad mood,” sleep disturbances, and cyberbullying victimization, practitioners can prioritize interventions that specifically target these areas to disrupt the reinforcing cycles between these psychological factors.

For middle schools, these findings suggest several actionable strategies. First, schools should implement early screening programmes to identify students exhibiting signs of depression, particularly those with a “sad mood”. This can be achieved through regular mental health check-ups and the use of validated screening tools. The identified students can then be referred to school counselors or mental health professionals for further assessment and support. Second, improving sleep quality among students should be a priority. Schools can collaborate with parents to educate students about the importance of sleep hygiene and to establish routines that promote adequate sleep. Third, addressing cyberbullying is crucial. Schools should develop comprehensive digital citizenship programmes that educate students about the responsible use of technology and the consequences of cyberbullying. Such programmes can be integrated into the curriculum and be reinforced through school-wide campaigns and awareness events. Additionally, schools should establish clear policies and procedures for reporting and addressing cyberbullying incidents, ensuring that victims feel supported and that perpetrators are held accountable. Fourth, managing fear-based anxiety symptoms through therapeutic approaches such as exposure-based interventions can help prevent aggressive responses. School counselors can offer group or individual therapy sessions that focus on building students’ resilience and coping skills.

CONCLUSION

This study underscores the complex interplay among internalizing problems (depression and anxiety), externalizing problems (aggression), and PBV in middle school students. Depression, particularly a “sad mood”, and sleep disturbances significantly influence both PBV and aggression, highlighting their pivotal role in the cycle of these issues. Cyberbullying victimization also emerges as a critical factor affecting mental health. These findings suggest that early intervention targeting these key elements could effectively mitigate the occurrence of PBV, depression, anxiety, and aggression. Schools should implement comprehensive strategies, including early screening for depression, sleep education, and digital citizenship programmes, to foster a supportive and inclusive environment. Addressing these issues holistically can enhance students’ mental well-being and promote a positive school climate.