Date of Award

Spring 2026

Document Type

Open Access Dissertation

Department

School of Information Science

First Advisor

Vanessa Kitzie

Abstract

Content moderation can reinforce stigma and limit access to much-needed information, including information on nonsuicidal self-injury. Despite its prevalence, the behavior remains highly stigmatized, prompting individuals to turn to social media platforms like TikTok for social support and harm reduction information. However, as algorithms are known to disproportionately moderate stigmatized content, users must navigate a careful balance between visibility and concealment, altering their information practices to avoid detection. Using an explanatory sequential mixed methods design, this study explored how users navigate content moderation, negotiate visibility, and adapt their information practices in anticipation of and in response to moderation to create and use nonsuicidal self-injury information on TikTok. First, an analysis of 973 nonsuicidal self-injury videos demonstrated that specific content characteristics, such as signaling, trigger warnings, and addiction language, are significantly associated with content moderation, while user engagement does not play a significant role. These findings suggest that TikTok’s moderation systems rely heavily on surface-level, machine-readable cues and may struggle to account for context. Semi-structured interviews with 21 TikTok users offered insights into the lived experiences behind these associations, showing how moderation directly intersects with users’ motivations and information practices. Findings revealed that users’ motivations to create and use nonsuicidal self-injury information are deeply social and emotional, with users primarily motivated to build community and de-stigmatize mental health. Findings also highlighted an interplay folk theories and information practices, where users’ folk theories shape the information creation and use practices they employ to evade moderation while the outcomes of these practices subsequently reshape their folk theories, creating an iterative cycle. These findings advance theoretical understandings of information marginalization by illustrating how platform governance creates new forms of marginalization, which in turn drives users to adapt their information practices. Thus, rather than framing users’ information practices as defective or deviant, findings demonstrate that these practices are rational responses to content moderation. Ultimately, this study underscores the need to re-examine how content moderation disproportionately affects stigmatized mental health communities. In doing so, this study offers practical implications for trauma-informed content moderation, including human-AI collaboration and context-sensitive moderation, that attend to the ways users express and process their experiences with mental health on social media.

Rights

© 2026, Valerie Angela Vera

Download

Share

COinS