In post-conflict societies where mental health infrastructure is sparse and trauma exposure is already elevated, digital access may be amplifying harm in ways that public health systems are wholly unprepared to address. A cross-sectional survey from Juba, South Sudan — among the first of its kind from a fragile African state — puts hard numbers on a risk landscape that has largely been invisible to researchers.
Among 528 children aged 8–17 surveyed via face-to-face household interviews, 69% had engaged in at least one risky online behavior, defined as a composite of five interpersonal conduct types, and 23% reported lifetime online victimization encompassing harassment, bullying, or abuse. The analysis revealed a pronounced bidirectional relationship between risky behavior and victimization — each strongly predicting the other in both unadjusted and adjusted logistic regression models. Secondary school enrollment was associated with dramatically elevated odds of risky online behavior (adjusted odds ratio: 6.45), and male gender emerged as a consistent predictor. Notably, educational internet use was protective against victimization (AOR: 0.44), while moderate daily use of one to five hours was associated with increased victimization odds (AOR: 1.63).
These findings carry implications that extend well beyond South Sudan. The global discourse on children's digital safety has been dominated by data from high-income countries with robust regulatory frameworks and accessible mental health support. Evidence from fragile or post-conflict settings challenges the assumption that digital access is uniformly empowering, particularly when oversight structures are absent. The bidirectional victimization-behavior relationship suggests a reinforcing cycle that mirrors offline trauma dynamics — a finding consistent with adverse childhood experience research, though causal direction cannot be established in a cross-sectional design. The protective effect of educational internet use is a meaningful policy signal, but sampling limitations — purposive rather than random — constrain generalizability. This is incremental but foundational evidence for a severely underrepresented population.
