University life now unfolds across lecture halls, learning platforms, group chats and social networks. That overlap creates opportunities for collaboration, but it also means harmful online behaviour can follow students directly into their academic lives. A newly published study of 1,862 undergraduates in Saudi Arabia offers unusually detailed evidence on the behavioural factors associated with cyber harassment and how those patterns relate to students’ perceived academic difficulties.
Published in Scientific Reports, the research applies the theory of planned behavior to cyber harassment, shifting attention from victimisation alone toward the attitudes, social pressures and perceived control associated with students’ own self-reported engagement. The analysis found that cyber harassment engagement was positively associated with perceived academic performance impairment, with a standardised coefficient of 0.38. The relationship was statistically significant at p < 0.001.
A behavioural model of online harassment
The researchers surveyed undergraduate students from five Saudi public universities: King Abdulaziz University, King Saud University, King Faisal University, Jazan University and the University of Hail. Proportional stratified sampling was used to determine representation across the institutions, while the questionnaire was distributed through university communication channels and student networks.
After incomplete and ineligible responses were excluded, 1,862 valid questionnaires remained. Women represented 62.4% of the sample, or 1,162 respondents, while men accounted for 37.6%, or 700. Most participants were aged 18 to 27, a group comprising 92.1% of the sample. Nearly three quarters, 73.5%, were unmarried.
The study was cross-sectional, meaning all of the key measures were collected at one point in time. Participants responded to five-point Likert-scale items covering technology awareness, social media influence, cybersecurity knowledge, attitudes toward cyber harassment, subjective norms, perceived behavioural control, cyber harassment engagement and academic performance impairment. The Arabic survey was back-translated against the English version, and the questionnaire was pre-tested before the main data collection.
Importantly, cyber harassment engagement was not an externally observed record of misconduct. It was a self-reported construct capturing participation or behavioural tendency toward online conduct intended to harass, threaten or harm others. Academic impairment was also perceptual rather than an objective measure such as grades or examination results. It captured students’ reported academic disruption associated with cyber harassment involvement or exposure.
Social pressure and perceived control stood out
The team analysed the data using structural equation modelling in AMOS 21. The measurement model showed acceptable fit, with a Comparative Fit Index of 0.942 and a root mean square error of approximation of 0.045. The subsequent structural model produced a CFI of 0.944 and RMSEA of 0.046, again indicating that the proposed model represented the observed covariance structure reasonably well.
Within that model, attitudes toward cyber harassment were negatively associated with engagement, while subjective norms and perceived behavioural control were positively associated with it. In practical terms, the pattern suggests that personal evaluations of harassment, perceptions of social expectations and the perceived ease of carrying out harmful online behaviour all mattered within the model.
The perceived-control result is particularly important to interpret carefully. In this context, greater control does not mean better self-control. It reflects how easy or difficult a person believes performing the behaviour would be. The authors argue that online anonymity, reduced immediate accountability and the technical ease of digital communication may contribute to a sense that harassment can be carried out with relatively few obstacles.
The three central theory-of-planned-behavior variables together explained 77% of the variance in the study’s cyber harassment engagement construct. The wider model also incorporated technology awareness, social media influence and cybersecurity knowledge as antecedents of attitudes, norms and perceived behavioural control. The respective R² values for subjective norm, attitude and perceived behavioural control were 0.78, 0.85 and 0.81.
The academic link was meaningful but not causal
Cyber harassment engagement was significantly associated with perceived academic performance impairment, with β = 0.38 and p < 0.001. In the structural model, harassment engagement explained 15% of the variance in academic impairment.
That is substantial enough to merit attention, but it should not be translated into a claim that harassment caused poorer academic performance. The study cannot establish which factor came first, and both variables were reported by the same participants. Students experiencing academic difficulties could differ in other ways that are also related to their online behaviour, while unmeasured social or psychological factors may contribute to both.
The distinction is especially important because the academic outcome was not based on institutional records. The analysis therefore speaks to perceived academic impairment, not a measured fall in marks. Even so, the association provides a useful basis for longitudinal work that could track online behaviour and academic outcomes over time.
Gender changed one social relationship, but not the whole model
The researchers also tested whether gender, marital status and geographical region altered selected relationships in the model. Multi-group structural equation modelling compared constrained and unconstrained models, with a chi-square difference test used to identify significant moderation.
Gender significantly moderated only the relationship between subjective norms and cyber harassment engagement. It did not significantly alter the links from attitude or perceived behavioural control to engagement. Neither marital status nor geographical region significantly moderated the tested relationships.
This matters for intervention design because it suggests that social expectations may operate differently across gender groups even when other behavioural relationships are more stable. At the same time, the absence of regional moderation in this sample cautions against assuming that every digital-safety programme needs a fundamentally different behavioural model for each part of the country.
What universities can take from the findings
The results point beyond awareness campaigns that simply tell students cyber harassment is harmful. If perceived peer norms and perceived ease of acting online are associated with engagement, prevention may also need to address the social environment around digital behaviour and the mechanisms that make misconduct feel easy or consequence-free.
Universities can influence that environment through clearer behavioural expectations, visible reporting channels, consistent consequences, digital citizenship education and support structures that make harmful norms less likely to become socially reinforced. The evidence does not prove that any particular intervention will reduce harassment, but it helps identify behavioural pathways worth testing.
The study also illustrates why cybersecurity knowledge alone should not be treated as synonymous with healthy online conduct. Technical knowledge sits within a broader behavioural system involving attitudes, social influence and perceived capability. Knowing how digital systems work does not automatically determine how people use them.
Important limits remain
The large sample and use of five public universities strengthen the analysis, as do the reliability, validity and model-fit checks. The researchers also tested common method bias with a single-factor model, which fitted the data poorly compared with the hypothesised measurement model.
Several limitations nevertheless narrow what can be concluded. The cross-sectional design prevents causal inference. Self-reported harassment is vulnerable to social desirability bias, particularly for behaviour that respondents may be reluctant to admit. The survey link was distributed through university channels and student networks, so the researchers could not determine precisely how many students received or opened it and therefore could not calculate a response rate.
The sample was also limited to public universities. Private institutions, students outside higher education and other national settings may show different behavioural patterns. Future longitudinal research could test whether changes in norms, attitudes or perceived control precede changes in harassment and whether those changes are followed by objectively measured academic outcomes.
For now, the strongest conclusion is narrower but still useful: among these 1,862 Saudi undergraduates, cyber harassment engagement sat within a measurable network of attitudes, perceived social expectations and behavioural control, and higher engagement was associated with greater perceived academic impairment. Understanding the social architecture around harmful online behaviour may therefore be as important as understanding the technology through which it occurs.
Source Information
Study: Cyber harassment engagement and its impact on academic performance impairment: a theory of planned behavior approach
Authors: Fahad Abdullah Moafa, Kamsuriah Ahmad, Muaadh Mukred, Mikkay Wong Ei Leen, Ali Safaa Sadiq and colleagues
Journal: Scientific Reports
Year: 2026
Published: 28 September 2026
DOI: 10.1038/s41598-026-63932-y
