Adolescent students from six high schools in Tongliao City, Inner Mongolia, participated in a questionnaire survey. A total of 826 questionnaires were distributed, and 125 were excluded for insufficient responses, resulting in 701 valid participants, yielding an effective response rate of 84.7% [this response rate is meant to reflect valid participants/eligible participants (i.e., all students who attended the participating classrooms)]. Details of the valid participants are presented in Table 1. Participants were from several schools in Naihanqi of Tongliao City, with 550 participants (78.5%) from rural areas and 133 participants (19.0%) from urban areas. Among the participants, 187 (26.7%) held class leadership positions, while 493 (70.3%) did not. Also, 225 participants (32.1%) were only children, while 464 (66.2%) were not. In terms of family structure, 563 participants (80.3%) came from intact families, 67 (9.6%) came from single-parent families, and 56 (8.0%) came from blended families. Females constituted 51.16%, while males comprised 48.84% of all participants. As for grade levels, there were 63 participants (9.0%) in the first year of junior high school, with an average age of about 13; 118 participants (16.8%) in the second year, with an average age of 14; 144 participants (20.5%) in the third year, with an average age of 15; 150 participants (21.4%) in the first year of senior high school, with an average age of 16; 143 participants (20.4%) in the second year, with an average age of around 17; and 74 participants (10.6%) in the third year, with an average age of about 18. Data collection and entry for this study ranged from November 2023 to January 2024.

Inclusion Criteria: The participants must be between the ages of 12 and 18, properly understand the contents of the questionnaire, and be willing to cooperate with the survey. Additionally, consent must be obtained from both the participants and their guardians.

Exclusion Criteria: Exclusion criteria include the presence of neurological or other mental disorders, substance or alcohol dependency, conditions such as epilepsy, brain injury, or intellectual disabilities, recent exposure to significant stressful events, hearing, speech, or other communication impairments that affect everyday interaction, and an inability to cooperate with the surveyors.

This study was approved by the Research Ethics Committee of Inner Mongolia University for Nationalities. After establishing contact with teachers from two middle schools in Naiman Banner, Tongliao City, the researcher introduced the purpose and content of the study. After acquiring their consent, the teachers in the local school used the paper version of the questionnaire to conduct the whole class test in their spare time, and the students completed the questionnaire anonymously. Written informed consent and assent forms were acquired from the participants and their legal guardians/relatives. Participation was voluntary, and confidentiality was guaranteed. The data collectors consisted of trained researchers who ensured the standardization of the data collection process. Completing the self-report questionnaire took approximately 30 min.

Initially, descriptive statistics and correlation analyses were conducted using SPSS 26.0. The next step involved performing regression analyses to assess potential violations of the underlying assumptions. A hypothesis was considered valid if the 95% confidence interval (CI) did not include zero. Subsequently, the mediation and moderated mediation models proposed by Hayes (2013) were applied using SPSS Models 4 and 15. A bootstrap sample of 5,000 was drawn from the data, and the 95% bootstrap confidence interval (CI) was calculated. Finally, gender and grade were included as covariates in the analysis.

Table 2 presents the means, standard deviations, and correlation coefficients for the variables examined in this study. The analysis revealed significant correlations among all variable pairs: students’ experiences of bullying, alexithymia, friendship quality, and non-suicidal self-injury (NSSI). Notably, friendship quality was negatively correlated with NSSI, students’ experiences of bullying, and alexithymia. This suggests that higher friendship quality is associated with a lower likelihood of engaging in NSSI, while lower friendship quality is linked to an increased risk of bullying victimization. Additionally, individuals with higher friendship quality exhibited lower levels of alexithymia. Conversely, significant positive correlations were observed between the other variables. Specifically, more frequent experiences of school bullying were associated with more severe alexithymia symptoms, and a greater degree of bullying experiences was linked to more frequent engagement in NSSI behaviors. Similarly, higher levels of alexithymia were associated with more severe NSSI scores. These findings provide empirical support for the hypotheses tested in the subsequent analysis.

According to Table 3, an F-test was conducted to evaluate the model, and the results indicated model validity (F = 34.01, p = 0.000 < 0.001). The total regression coefficient for student bullying was 2.67 (t = 5.83, p = 0.000 < 0.001), suggesting that student bullying has a significant positive impact on non-suicidal self-injury behaviors.

The mediation model was tested using SPSS Process Model 4, with gender and grade as control variables. The analysis revealed a significant positive association between students’ experiences of bullying and NSSI (β = 2.11, p < 0.001) (Table 4). As shown in Table 4, when alexithymia was added to the model, a significant positive correlation between students’ experiences of bullying and alexithymia was observed (β = 4.43, p < 0.001), and alexithymia was also positively associated with NSSI (β = 0.12, p < 0.001). Furthermore, the Bootstrapping test results, presented in Table 5, indicated that the 95% confidence intervals for both the indirect effect of alexithymia and the direct effect of students’ experiences of bullying on NSSI did not include zero ([0.00, 1.23], [0.22, 0.94]). Meanwhile, the mediating effect was 0.53, accounting for 20.08% of the total effect. These findings suggest that alexithymia partially mediates the relationship between students’ experiences of bullying and NSSI. Collectively, these results provide empirical support for Hypotheses 1 and 2.

The moderated mediation model was tested using SPSS Process Model 15. As shown in Table 6, the interaction term between students’ experiences of bullying and friendship quality had a significant direct effect on non-suicidal self-injury (NSSI) (β = −0.07, p < 0.05), and the interaction term between alexithymia and friendship quality also significantly influenced NSSI (β = −0.003, p < 0.01).

A simple slope analysis was conducted to explore the nature of these interactions. Participants were divided into two groups based on the level of moderation (friendship quality): low (M – SD) and high (M + SD). Among individuals with low friendship quality, a positive correlation between alexithymia and NSSI was observed (simple slope = 0.17, p < 0.001). In contrast, although the association remained significant for individuals with high friendship quality, it was weaker (simple slope = 0.08, p < 0.001). These findings suggest that higher levels of friendship quality attenuate the relationship between alexithymia and NSSI. These results support Hypothesis 3, confirming that friendship quality moderates the relationship between alexithymia and NSSI.

Similarly, when participants were divided into two groups based on friendship quality (low vs. high), a positive correlation between students’ experiences of bullying and NSSI was observed among individuals with low friendship quality (simple slope = 2.15, p < 0.001). However, for individuals with high friendship quality, this correlation, while still significant, was weaker (simple slope = 0.51, p < 0.001). These results indicate that higher levels of friendship quality diminish the strength of the relationship between students’ experiences of bullying and NSSI. Collectively, these findings support Hypothesis 4, confirming that friendship quality moderates the relationship between students’ experiences of bullying and NSSI.

This study demonstrates that students’ experiences of bullying are a direct predictor of NSSI behaviors, aligning with previous research indicating that students who experience bullying are at a heightened risk of engaging in NSSI (Li et al., 2020). When individuals encounter stressful situations, they cannot resolve or adapt. They may resort to maladaptive coping mechanisms to alleviate the emotional distress caused by these experiences. According to the general stress model, individuals exposed to environmental stressors like bullying and unable to modify their behaviors to cope with these stressors experience a subjective sense of oppression. This heightened distress increases their likelihood of engaging in extreme behaviors, such as NSSI or avoidance, to manage or escape the negative emotions induced by stress (Van Geel et al., 2015). Consequently, adolescents who experience bullying are more likely to develop NSSI behaviors compared to their non-bullied peers.

The findings of this study indicate that students’ experiences of bullying indirectly affect adolescents’ NSSI behaviors through alexithymia, thereby increasing the risk of NSSI. This result corroborates the work of Guo et al. (2024), who found that alexithymia mediates the relationship between bullying victimization and adolescent NSSI. Specifically, bullied students tend to exhibit higher levels of alexithymia compared to those who are not, and individuals with higher alexithymia levels are more likely to engage in NSSI behaviors (Guo et al., 2024; Wen et al., 2024).

On the one hand, victims of bullying often experience fragmented memories and a narrowed focus of attention (van der Kolk and Fisler, 1995), making it difficult for them to fully utilize their cognitive resources to process emotional states and physiological responses (Lischke et al., 2022). This impairment in emotional processing hinders their ability to recognize and express their own and others’ emotions, further preventing them from regulating traumatic feelings and exacerbating emotional dysregulation (Cowie and Berdondini, 2002). As a result, bullied students are more likely to develop alexithymia. On the other hand, the chronic exposure to trauma, such as bullying, generates persistent negative emotions, which adolescents may attempt to manage through maladaptive coping mechanisms like NSSI. This pattern of negative reinforcement exacerbates the cycle of emotional dysregulation. Therefore, individuals with alexithymia are more likely to engage in NSSI behaviors as a way to cope with emotional distress (Hasking, 2016). Empirical studies also support the notion that alexithymia is a direct predictor of NSSI (Zhang et al., 2023), making it one of the most significant predictors of NSSI (Cameron et al., 2014; Taylor and Bagby, 2004; Swiller, 1988).

Furthermore, these findings align with the social mechanism of alexithymia, which suggests that negative social experiences in childhood can disrupt an individual’s psychological well-being and reduce their capacity for social interaction (Honkalampi et al., 2004). Consequently, individuals who have experienced bullying may have impaired emotional regulation and self-evaluation processes, which contribute to alexithymia and significantly predict the risk of NSSI in adolescents (Guo et al., 2024). Severe alexithymia was associated with exacerbated NSSI following bullying, though this effect may be buffered by high friendship quality in certain individuals.

Finally, the study found that friendship quality moderates the relationship between alexithymia and NSSI.

Specifically, individuals with higher-quality friendships exhibited lower NSSI scores than those with lower-quality friendships and higher levels of alexithymia. These findings suggest that friendship quality may act as a protective factor against alexithymia. Previous research similarly demonstrated that friendship quality moderates the impact of negative emotions on NSSI (Wang et al., 2021), in line with the Social Buffering Model. According to this model, friendship quality, as a form of social support, can mitigate the negative effects of adverse experiences (Cohen and Wills, 1985). Therefore, when adolescents experience higher levels of friendship quality, it can alleviate negative emotions and reduce the likelihood of maladaptive behaviors, including NSSI (Kim, 2015; Sinanan and Gomes, 2020; Woods et al., 2009). Empirical studies also support this notion, showing that lower-quality friendships are associated with more pronounced alexithymia symptoms (Peng and Huang, 2024), while individuals with higher-quality friendships tend to engage in fewer NSSI behaviors (Eggermont et al., 2021). Moreover, NSSI is often considered an unhealthy emotion regulation strategy and maladaptive coping behavior (Klonsky, 2009). Adolescents with high-quality friendships are more likely to manage psychological distress through social support and communication rather than resorting to NSSI.

Additionally, the study found that friendship quality also moderates the relationship between students’ experiences of bullying and NSSI. Specifically, individuals with higher-quality friendships exhibited lower NSSI scores than those with lower-quality friendships. These findings are consistent with those of Wang et al. (2021), who showed that friendship quality moderated the relationship between cyberbullying and NSSI. Most studies consistently indicate that friendship quality moderates the relationship between adverse experiences and severe outcomes (Bernasco et al., 2022; Li et al., 2023; Li et al., 2022). High-quality friendships are characterized by low levels of conflict/arguments and high levels of positive features, including prosocial behaviors, loyalty, intimacy, and self-esteem support. Notably, the constructive subdimensions of such friendships can effectively mitigate the severe consequences associated with bullying victimization (Kloo, 2025). Furthermore, the conclusions of this study align with the Social Buffering Model, which suggests that individuals with higher-quality friendships are better equipped to cope with challenges through emotional support and problem-solving strategies derived from interpersonal interactions. This helps mitigate the adverse effects of bullying. Therefore, in similar risk environments, individuals with high-quality friendships are less likely to engage in harmful behaviors (Cohen and Wills, 1985).

Moreover, according to the General Strain Theory, recent exposure to other major stressful events (included in the exclusion criteria) may also have a potential impact on NSSI. Additionally, studies have shown that substance or alcohol dependency is associated with NSSI behaviors, while speech or other communication impairments could significantly confound data analysis. Therefore, these factors were deemed appropriate exclusion criteria to enhance the robustness of the study findings.

This study contributes to the social mechanism perspective of alexithymia, providing valuable insights into its role as a significant predictor of NSSI in the context of student bullying. Specifically, the results show that students’ experiences of bullying indirectly affect adolescent NSSI through the mediating role of alexithymia. Furthermore, our findings deepen the understanding of general stress theory, the social mechanism perspective of alexithymia, the cognitive-emotional model of NSSI, and social buffering theory. They also highlight the moderating role of friendship quality in both the relationship between students’ experiences of bullying and NSSI and the relationship between alexithymia and NSSI.

Firstly, Given that students’ experiences of bullying are significant predictors of non-suicidal self-injury (NSSI) behavior, it is essential for schools to address bullying incidents swiftly and per established protocols. Such incidents should be treated with the highest priority, ensuring prompt intervention. In cases where necessary, parental involvement should be sought to provide guidance and support, facilitating a timely and effective resolution. On the other hand, in the absence of bullying incidents, schools should conduct regular seminars and awareness campaigns for students and parents to promote self-protection, raise parental awareness, and establish and implement robust prevention and intervention policies (Guo et al., 2024). These policies should include the development of a comprehensive monitoring and disciplinary system (Li et al., 2020), fostering close collaboration between school staff and parents to form a supportive school-family partnership. This partnership should focus on closely monitoring students’ psychological well-being and behavioral indicators, ensuring the early identification of potential bullying victims and the implementation of targeted interventions. By addressing bullying early, these measures will help reduce bullying incidents and provide a foundation for lowering the incidence of NSSI. More importantly, the school-family partnership should actively involve students, encouraging them to develop effective coping strategies. Influenced by a supportive campus environment and related activities, students can enhance their ability to manage stress, navigate challenges, and learn to protect themselves and their peers from harm.

Secondly, the present study highlights the significant role of alexithymia in the relationship between student students’ experiences of bullying and non-suicidal self-injury. Given this, parents and educators need to promote the development of emotional regulation strategies and social skills in students, particularly those exhibiting symptoms of alexithymia, as they may face greater challenges in managing their emotional responses to adverse experiences. Schools and families can facilitate this process by organizing group activities, such as team-based games, to enhance students’ ability to recognize, express, and regulate their emotions (Guo et al., 2024; Zhang et al., 2023). Using appropriate language, these activities should encourage students to articulate their inner experiences and feelings openly. Such interventions may help alleviate the negative symptoms commonly associated with alexithymia. In addition, students should be encouraged to engage in regular self-reflection and external communication to improve their emotional awareness and coping skills. Parents, in turn, must prioritize their children’s emotional development by fostering an environment conducive to emotional expression and communication. They should stay attuned to their children’s emotional states and guide them in developing effective coping strategies to enhance emotional resilience and well-being.

Finally, considering the pivotal role of friendship quality in moderating the relationship between students’ experiences of bullying and NSSI, schools need to create opportunities for positive peer interactions at key moments, fostering mutual support and expanding students’ social networks. For example, schools could organize friendship-building activities across different classes to establish and strengthen positive friendships among middle school students. These initiatives would promote the formation of supportive social connections and enhance the overall quality of their relationships. Furthermore, schools should periodically rearrange class groups to increase students’ opportunities to meet new peers and form diverse social connections (Cuadros and Berger, 2016). Such interactions help students develop essential social skills and significantly improve the quality of their friendships. In parallel, students should be actively encouraged to develop a deeper understanding of friendship dynamics and social skills, focusing on cultivating interpersonal abilities that improve the quality of their relationships. Students can build positive, supportive friendships that promote mutual growth by intentionally fostering these skills. It is also important to note that the quality of students’ friendships can be influenced by family dynamics and relational patterns within the home environment. Consequently, parents should actively nurture their children’s social skills, create a supportive atmosphere that encourages friendship-building, and offer emotional support, all of which contribute to developing their children’s social competence and emotional well-being.

Although this study makes significant contributions by exploring key issues and uncovering the underlying patterns of essential phenomena, it also reveals certain limitations that should be addressed in future research. These limitations provide opportunities to refine the findings and enhance their robustness. First, the study’s participant pool was restricted to rural middle school students, which may limit the generalizability of the results. Future research should, therefore, consider including urban middle school students in the sample and establish a control group for comparative analysis. Such an approach would broaden the findings’ scope and improve the conclusions’ comprehensiveness and validity.

Second, this study utilized a cross-sectional design, which, while effective for examining relationships between variables at a single point in time, does not capture dynamic changes or long-term developmental trends. Future studies should incorporate longitudinal tracking methods to better understand and validate the complex relationships between variables. This would involve conducting long-term follow-up observations and data analysis to capture the evolving nature of the variables over time, thereby enhancing the reliability and stability of the findings. Finally, this study primarily focused on the theoretical discussion of mediating and moderating effects, providing valuable conceptual insights for the field. However, its practical applications and implications for real-world interventions remain limited. Future research should expand its scope by incorporating intervention studies to address this gap. By designing and implementing targeted intervention strategies, future studies could assess the feasibility and impact of the theoretical findings, providing more concrete and actionable guidance for real-life applications. Moreover, such research would offer valuable lessons and insights to inform future intervention efforts. Finally, negative friendship traits, including conflict-ridden relationships, can lead to various adverse outcomes, thereby increasing the risk of victimization. However, this study lacks an in-depth examination of the relationship between specific subdimensions of friendship quality and the dependent variables. Future research exploring these associations could yield more profound insights.