Exploring the mediating roles of bullying, school fighting, and media literacy in the relationship between social-emotional competence and cyberbullying among adolescents | BMC Psychology

Sampling and design

This study is classified as applied research and employs a non-experimental design, along with correlational analysis through path modeling, for data collection and analysis. It is framed within a cross-sectional framework to explore associations among variables rather than infer causal relationships. The target population for this research comprises all female high school students in three cities in China (Beijing, Chengdu, and Xi’an) and one city in Iran (Shiraz) during the 2023–2024 academic year. These locations were selected to reflect both cultural diversity and regional differences in digital engagement, as well as the prevalence of cyberbullying. China and Iran represent two distinct socio-cultural contexts where internet access, educational policy, and adolescent digital behavior may vary, thereby enhancing the cross-cultural relevance of the findings.

A total of 564 students were selected using a multi-stage cluster sampling technique. The initial stage involved randomly selecting schools within each city, followed by class-level sampling. Inclusion criteria required participants to be currently enrolled in high school, aged between 14 and 17, and active users of social media. Students with diagnosed cognitive impairments or those who had not used digital platforms in the previous six months were excluded from the study to maintain the validity of responses in media-related measures.

Demographic characteristics of the sample revealed that all participants were female, with an average age of 15.8 years (SD = 0.9). The socio-economic status (SES) of participants was assessed through a brief survey capturing parental education and household income. Approximately 42% of the sample came from middle-income households, 38% from low-income households, and 20% from high-income families. While the sample size is sufficient for structural equation modeling, the use of an all-female sample limits the generalizability of findings to male or mixed-gender populations. Given that gender differences in cyberbullying behavior are well-documented [e.g., 26], the results may reflect gender-specific trends in cyber victimization and emotional regulation. Future studies should replicate this research with more diverse gender compositions to explore potential moderation effects.

Research instruments

Social-Emotional Competence Questionnaire (SECQ): Developed by Zhou and Ee [37], this questionnaire comprises 25 items that assess five dimensions of social-emotional competence (SEC) on a five-point Likert scale: self-awareness, social awareness, self-management, relationship management, and responsible decision-making. The validity of this instrument has been established through factor analysis in prior studies conducted by its creators. Furthermore, the current research calculated its reliability at 0.88 using Cronbach’s alpha.

Illinois bully scale (IBS)

This scale, developed by Espelage and Holt [36], consists of 18 items designed to evaluate behaviors related to bullying, fighting, and victimization, using a five-point Likert scale. The overall reliability coefficient for the scale was determined to be 0.87, with subscale reliabilities of 0.71 for victimization, 0.77 for bullying behaviors, and 0.76 for fighting behaviors. Validity was confirmed through factor analysis as well. In this study, the reliability coefficients for the bullying and fighting components were found to be 0.69 and 0.72, respectively, as determined by Cronbach’s alpha. While the bullying subscale fell slightly below the conventional threshold of 0.70, it was retained based on prior validation and theoretical relevance. This limitation is acknowledged as a minor compromise in measurement consistency that does not significantly undermine overall interpretability.

Media literacy questionnaire

This instrument, developed and standardized by Simons et al. [29], includes 12 items that assess media literacy among adolescents. It encompasses three components: media engagement, media use, and media perception. In the original research conducted by its authors, reliabilities were reported at 0.63, 0.70, and 0.78, respectively. They also indicated satisfactory validity for this instrument. In the present study, the reliability of this questionnaire was determined to be 0.88 based on Cronbach’s alpha. The notable improvement in reliability may reflect cultural adaptation or consistent interpretation among the target population, although item redundancy could be a potential factor to monitor in future applications.

Cyberbullying victimization experience questionnaire

Designed by Patchin and Hinduja [38], this questionnaire aims to assess experiences related to cyberbullying or cyber victimization among adolescents. Scoring is based on a five-point Likert scale format. It features two factors: (1) cyber victimization and (2) cyberbullying; each consists of nine questions. Research conducted by the developers demonstrated that this instrument possesses adequate validity. Additionally, in this study, the Cronbach’s alpha coefficient for both factors exceeded 0.89, indicating strong reliability overall. For the current research context, the reliability coefficient for the cyberbullying factor was determined to be 0.93 using Cronbach’s alpha. While this value suggests excellent internal consistency, it may also indicate item redundancy, warranting further analysis through item-total correlations or exploratory factor analysis in future research.

Procedure

The data collection procedure involved administering four established questionnaires to the participants. These surveys were administered in a controlled classroom environment to ensure uniformity and consistency in their presentation. The participants completed the questionnaires under the supervision of trained research assistants who were available to answer any questions or clarify instructions during the process. To minimize bias and ensure the reliability of the responses, the students were assured of confidentiality and anonymity regarding their answers. The entire data collection process spanned two weeks, with all instruments administered simultaneously to allow for the measurement of the relevant constructs at once.

Ethical approval

for this study was obtained from both the university’s research ethics committee and the local education authorities in each participating region. Informed consent was obtained from school administrators, parents, and participants themselves. Participants were explicitly informed about the voluntary nature of the study, their right to withdraw at any time, and the confidentiality of their responses. Data were anonymized before analysis to protect participants’ identities and ensure compliance with ethical research standards in cross-cultural studies involving minors.

Once the data collection was completed, the responses were compiled, and the initial dataset underwent data cleaning. Incomplete or inconsistent responses were excluded to maintain the integrity of the data. The final sample retained only those responses with at least 95% completion across all survey items.

Data analysis

The data analysis was conducted using Structural Equation Modeling (SEM) through path analysis, a method well-suited for examining the direct and indirect relationships between multiple variables simultaneously. This method allowed for testing the hypothesized relationships between social-emotional competence, bullying behaviors, fighting, media literacy, and cyberbullying victimization. Before conducting the SEM analysis, the data were checked for normality, multicollinearity, and outliers to ensure the validity of the statistical techniques applied.

Descriptive statistics were first calculated for each variable, including means, standard deviations, and reliability coefficients, to summarize the data and assess internal consistency. Cronbach’s alpha was used to evaluate the reliability of the instruments used in the study, and all variables exhibited acceptable internal consistency, with coefficients above 0.69. Next, a confirmatory factor analysis (CFA) was conducted to validate the measurement model of the study, ensuring that each instrument accurately measured its intended construct. Once the measurement model was validated, the hypothesized structural model was tested, and path coefficients were estimated.

Bootstrapping procedures were used to assess the significance of indirect effects, allowing for the evaluation of mediation pathways between social-emotional competence and cyberbullying through bullying behaviors, fighting, and media literacy. The model fit was assessed using multiple indices, including the Comparative Fit Index (CFI), the Root Mean Square Error of Approximation (RMSEA), and the Standardized Root Mean Square Residual (SRMR). These indices provided a comprehensive evaluation of the model’s goodness of fit, ensuring the robustness of the results.

All statistical analyses were conducted using SPSS and AMOS software. The significance level for all tests was set at p < 0.05, and the results were interpreted in the context of existing literature to conclude the relationships between the variables under investigation. Findings are discussed in light of the limitations posed by sample homogeneity and regional representation, with caution applied in extending conclusions to broader adolescent populations.

Findings

Table 1 displays the descriptive statistics for the examined sample, including the mean, standard deviation, skewness, and kurtosis, along with Cronbach’s alpha coefficient for the relevant variables. The data analysis was performed using SPSS-16 and Lisrel-10 software.

As indicated in Table 1, the values of skewness and kurtosis (falling between + 2 and − 2) for all research variables suggest that their distribution is normal. Consequently, path analysis can be utilized for data analysis. Pearson’s correlation coefficient was computed to assess the interrelationships among the variables. The findings are displayed as a correlation matrix in Table 2.

The matrix indicates that at a significance level of P < 0.01, social-emotional competence was significantly correlated with cyberbullying (r = -0.249). At the same time, bullying showed a significant relationship with cyberbullying (r = 0.492) and fighting (r = 0.444). Furthermore, at P < 0.01, social-emotional competence was significantly associated with bullying (r = -0.460) and media literacy (r = 0.336). Additionally, bullying was correlated considerably with fighting at P < 0.01 (r = 0.698).

Hypothesis testing and path analysis

This research developed a conceptual model to elucidate the structural relationships among variables grounded in existing theoretical and empirical literature. It was posited that cognitive abilities and perfectionism serve as mediators in the relationship between perceived parenting styles and obsessive beliefs. Path analysis employing the maximum likelihood method was utilized to evaluate the fit of the proposed models with the collected data and to estimate effect coefficients for each model. Descriptive statistics indicated that the distribution of all variables closely approximated normality; thus, the assumption regarding the normal distribution of data was satisfied. The adequacy of sample size was recognized as a crucial element in modelling. Bentler and Chou (1988), as referenced by Hooman (2009), assert that if the distribution of variables is normal and there are no outliers or missing values, the sample size can be reduced to five observations for each estimated parameter. Consequently, a sample size of 169 participants was deemed sufficient for this study. The bivariate correlation matrix for all variables exhibited an elliptical shape, suggesting the existence of linear relationships among them.

Table 3 presents the fit indices for the proposed model in relation to the observed data. As noted by Thompson (as cited in Meyers et al., 2012), a selection of overall fit indices—specifically the Chi-Square/df ratio, Normed Fit Index (NFI), Comparative Fit Index (CFI), and Root Mean Square Error of Approximation (RMSEA)—are among the most widely accepted indicators of structural model adequacy. In addition, the Goodness-of-Fit Index (GFI) was reported to provide further confirmation of model performance. Consequently, these indices were computed for the current study (see Table 3; Fig. 2). The Chi-Square/df value of 0.016, along with the RMSEA value of 0.0644, falls within the acceptable range for a good model fit. Moreover, the high values for NFI (0.967), CFI (0.986), and GFI (0.984) exceed the commonly recommended threshold of 0.90, indicating strong model fit. Overall, the combination of these indices suggests that the proposed research model provides a favorable and statistically acceptable representation of the data structure.

T-values in the Significance Test of the Proposed Research Hypothesis Model

Table 4 illustrates the direct and indirect effects in the final model, highlighting the complex relationships between social-emotional competence (SEC), bullying, school fighting, media literacy, and cyberbullying among Iranian adolescents. Among the direct effects, fighting demonstrated a significant positive association with cyberbullying (β = 0.21, SE = 0.06, t = 2.44, R² = 0.20), while bullying exhibited an even stronger positive influence (β = 0.38, SE = 0.08, t = 4.32). Additionally, media literacy had a modest but significant positive impact on cyberbullying (β = 0.14, SE = 0.05, t = 2.17).

Similarly, bullying had a significant influence on fighting, with a strong positive effect (β = 0.65, SE = 0.07, t = 10.91, R² = 0.42), indicating that bullying behaviors are closely linked to physical altercations in school settings. In contrast, SEC had a negative direct effect on bullying (β = -0.37, SE = 0.05, t = -5.04, R² = 0.13), suggesting that higher levels of SEC reduce the likelihood of bullying behaviors. Moreover, SEC positively influenced media literacy (β = 0.38, SE = 0.07, t = 5.18, R² = 0.14), indicating that adolescents with stronger social-emotional skills are more likely to develop media literacy.

For the indirect effects, SEC indirectly reduced fighting (β = -0.24, SE = 0.05, t = -4.59), mediated by its impact on bullying. Furthermore, SEC indirectly reduced cyberbullying (β = -0.13, SE = 0.04, t = -2.65) through pathways involving bullying and fighting. Finally, bullying indirectly increased cyberbullying (β = 0.51, SE = 0.05, t = 2.38), emphasizing its central role as a mediator in the model. These findings underscore the interplay between SEC, bullying, and media literacy in shaping adolescents’ behaviors related to fighting and cyberbullying.