Digital traces of child maltreatment: Investigating TikTok data donations and predicting depressive symptoms in adolescents

While the debate about social media and adolescent mental health is ongoing, there is growing consensus about exacerbated effects for vulnerable adolescents, e.g., after child maltreatment (CM). Existing research predominantly relies on self-reports, cross-sectional designs, and lacks analyses of specific social media activities. In a participatory digital data donation design, 129 adolescents (ages 13-18years, 31.8% exposed to CM) shared parts of their TikTok data archives capturing objective usage (e.g., videos viewed, posts, likes). Machine learning identified the average number of weekly searches as the most important TikTok behavior classifying CM status, followed by TikTok session length and the mean number of posts per week. Longitudinal analyses of identified TikTok behaviors with depressive symptoms revealed more followers and less posting activity as significant predictors of increased depression six months later. Findings will inform our understanding of how CM-exposed adolescents use TikTok differently from their peers and provide opportunities for targeted prevention.