The study arrives at a moment when artificial intelligence and digital technologies are transforming online communication at unprecedented speed. Smartphones, social media platforms, and instant messaging applications have created fresh arenas where hostility can spread rapidly, often anonymously and with little social disapproval. International research has repeatedly linked victimization to anxiety, depression, and in the worst cases suicidal thoughts, while the normalization of aggression online has been tied to broader social harms, including the perpetuation of power dynamics. Yet the academic understanding of what drives these behaviors remains fragmented, with demographic characteristics, peer reinforcement, platform anonymity, and content moderation policies typically examined in isolation rather than within a single analytical framework.

To close that gap, the research built on the ecological model of aggression, which holds that individual traits, the legal and policy environment, and the social environment all contribute to perpetration. The framework was reinforced by complementary theories: Social Learning Theory explains how aggressive behaviors are learned and reinforced through family and peer environments; Routine Activity Theory highlights how anonymity and the absence of effective guardianship create opportunities for aggression; Strain Theory links social frustration to aggressive coping; and the General Aggression Model connects repeated exposure to aggression with the strengthening of hostile cognitive scripts. Together, these perspectives allowed the study to treat cyberbullying as a potential precursor to cyber-violence rather than as a conceptually equivalent construct, a distinction that was empirically verified through discriminant validity procedures.

Jordan offered a distinctive setting for the investigation. The country has a fast-growing digital economy and a youthful, technology-oriented population that actively uses social media. At the same time, its legal framework, represented by the 2018 Cybercrime Law and ongoing policy changes, has not fully incorporated provisions targeting online harassment, leaving victims and offenders in a regulatory grey zone. The enforcement gap is striking: only about 12 percent of reported cyberbullying incidents result in formal legal action, and cybercrime cases rose between 2015 and 2022 despite the presence of legislation. Cultural norms around privacy, gender, and authority, along with a highly collectivistic society in which family reputation often outweighs individual expression, further shape how online aggression is expressed and perceived.

Methodologically, the study employed a quantitative design with a structured questionnaire whose 20 measurement items were adapted from well-established scales. The Arabic-language instrument was reviewed for linguistic clarity and conceptual equivalence, and content validity was assessed by four academic experts spanning digital sociology, clinical psychology, cyber law, and information systems. Responses were collected on a five-point Likert scale through both paper and online formats, using random sampling to reduce bias and monitoring demographic distribution throughout fieldwork. After excluding incomplete questionnaires, 472 valid responses were retained, a sample size exceeding the recommended respondent-to-item ratio for multivariate analysis. The study received ethical approval from the Research Ethics Committee of Al-Zaytoonah University of Jordan, and written informed consent was obtained from all participants.

The data were analyzed using partial least squares structural equation modeling in SmartPLS, a technique suited to prediction-oriented research because it maximizes the explained variance of endogenous constructs. The measurement model proved robust: Cronbach’s alpha coefficients ranged from 0.753 to 0.863, composite reliability values from 0.760 to 0.875, and average variance extracted between 0.580 and 0.708, all above accepted thresholds. Heterotrait-monotrait ratios remained below the conservative 0.85 cutoff, confirming adequate discriminant validity even between the closely related constructs of cyberbullying and cyber-violence. A full collinearity assessment with variance inflation factors ranging from 1.000 to 2.836 ruled out common method bias, and PLSpredict confirmed strong out-of-sample predictive relevance, with positive Q²predict values for both outcomes.

The structural model delivered a technically precise picture of the drivers involved. Individual characteristics, the legal and policy environment, cyberbullying, and the aggressive social environment together explained 43.6 percent of the variance in cyber-violence, while the predictors other than cyberbullying explained 46.5 percent of the variance in cyberbullying. Age showed a significant negative relationship with both outcomes, with path coefficients of −0.243 for cyber-violence and −0.275 for cyberbullying, indicating that younger individuals are markedly more likely to perpetrate online aggression. Gender showed a significant positive effect on both, with a larger coefficient for cyberbullying at 0.335, pointing to meaningful differences across gender groups. Educational level, by contrast, had no significant effect on either outcome, suggesting that formal education alone does not act as a protective factor in the Jordanian context.

The contextual factors showed differential effects that carry substantial policy weight. The aggressive social environment was positively associated with both cyberbullying and cyber-violence, with coefficients of 0.297 and 0.197 respectively, confirming its role as a key contextual driver of online aggression. The legal and policy environment, however, significantly deterred only cyber-violence, with a coefficient of −0.154, while its effect on cyberbullying was statistically insignificant. The study interprets this asymmetry through deterrence theory: formal sanctions are more effective against severe, recognizable, and reportable offences than against normalized behaviors that occur within routine peer interactions. Cyberbullying, being less often explicitly identified and defined, appears to fall below the threshold at which legal deterrence registers, whereas the greater visibility of enforcement and the severity of sanctions for violent acts make regulation more consequential there.

Perhaps the most consequential finding is the significant positive effect of cyberbullying on cyber-violence, with a path coefficient of 0.229. This supports an escalation-based understanding of online aggression: repeated involvement in cyberbullying may reinforce aggressive behavioral patterns, strengthen hostile cognitive scripts, and normalize hostile interactions, gradually increasing the propensity for more severe forms of cyber-violence, including threats, coercion, and harassment with offline consequences. The progression is not strictly linear, as prior research shows that strong self-regulation skills and supportive social environments can interrupt the pathway. Still, the evidence positions cyberbullying as a developmental stage rather than an endpoint, underscoring why early intervention matters before mild aggression hardens into something more dangerous.

The practical implications extend across policy, education, and platform design. Because the aggressive social climate exerts a stronger and more immediate influence than institutional rules, prevention efforts should focus on reducing the normalization of aggressive communication within peer and family settings, promoting constructive online interaction, and strengthening awareness of the consequences of digital hostility, all adapted to the Jordanian cultural context rather than relying on generic digital-safety approaches. Legal clarity and effective enforcement could help limit severe forms of online aggression, but sanctions alone are insufficient; they must be combined with social and educational interventions addressing peer norms. The significant effects of age and gender call for differentiated prevention strategies, with programs targeting younger people and adopting gender-sensitive approaches where appropriate. The authors acknowledge limitations, including the cross-sectional design, reliance on self-reported data, and the Jordanian cultural specificity of the findings, and recommend longitudinal research incorporating personality traits, digital literacy, online self-regulation, and social media usage patterns. What emerges is a clear message for a rapidly digitalizing world: cyber aggression is not an outcome of personal pathology alone but a multidimensional interaction between individuals, institutions, and the aggression-supportive norms woven into social environments.

Subject of Research: Cyberbullying and cyber-violence in Jordan are shaped by age, gender, social climate, and legal deterrence, with cyberbullying acting as a precursor to more severe cyber-violence.

Article Title: Explaining cyberbullying and cyber violence in the age of AI using individual and contextual factors in Jordan

Article References: Al-Omoush, K. S. (2026). Explaining cyberbullying and cyber violence in the age of AI using individual and contextual factors in Jordan. Discover Global Society, 4(1), Article 278. https://doi.org/10.1007/s44282-026-00625-x

Image Credits: AI Generated

DOI: 10.1007/s44282-026-00625-x

Keywords: cyberbullying, cyber-violence, Jordan, online aggression, social environment, legal policy, PLS-SEM, digital society, Explaining, cyber, violence, individual