Yu-Hsuan Fu
Background and Aim Debates about regulating social media to protect young people’s mental health have intensified internationally, yet policy discussions still often proceed without a clear account of how young people themselves define protection, evaluate acceptable trade-offs, and judge what counts as credible, fair, and workable regulation in practice. This thesis examines how adolescents and young adults in Aotearoa New Zealand define mental health protection in the context of social media regulation, how they evaluate the acceptability of possible regulatory responses, and how their reasoning can inform more credible and fair governance. Methods Using a qualitative design, I recruited 23 young people aged 16 to 25 from a range of gender and ethnic backgrounds, most of whom were students, to participate in focus groups that combined guided discussion, policy examples used as prompts, an individual Rule-Card activity, and a Follow-up Reflection Form. This staged design was intended to move beyond capturing broad opinions and instead elicit how participants translated values into workable proposals while reasoning through trade-offs, implementation constraints, and questions of responsibility. In doing so, it increased the analytic value of the data for examining feasibility, workarounds, privacy, equity, and governance reasoning in practice. Data from group discussions and reflection forms were analysed using reflexive thematic analysis informed by a critical realist perspective. Findings Participants did not approach regulation as a simple question of whether social media was good or bad, or whether young people should be online at all. Rather, they treated regulation as a problem of defining what mental health protection should mean under lived digital conditions. Harms were described as real and recurrent, especially where platform environments appeared to intensify compulsive engagement, erode agency over attention and time, disrupt sleep and daily functioning, or expose young people to disturbing or developmentally mistimed content that could linger psychologically or become normalised through circulation. At the same time, participants also identified valued benefits, including connection, learning, entertainment, identity exploration, access to support, and, for some, opportunity and livelihood. Protection was therefore usually reasoned through as reducing avoidable psychosocial cost by changing harmful conditions of use while preserving legitimate participation, rather than as a simple removal logic. A central finding is that acceptability was constructed not as a stable attitude of support or opposition, but as something regulatory proposals had to earn through credibility and fairness. Participants evaluated regulation through what this thesis interprets as a practical ethics of regulation: proposals were judged by whether they could plausibly reduce harm without introducing new risks, being easily bypassed, or distributing burdens unfairly. Feasibility sat at the centre of this reasoning. Restrictions vulnerable to workarounds, concealment, platform migration, or virtual private networks were often viewed as symbolic rather than protective. Age verification proposals involving identity documents or biometrics were judged disproportionate because they introduced privacy and security risks while offering uncertain protective value. Participants also tested credibility against bad-faith behaviour, for example where moderation could be gamed through labels or captions, and against implementation costs. Fairness was assessed through lived context, with participants questioning models relying heavily on parental gatekeeping, household digital literacy, or stable home conditions. When participants moved from evaluation to design, they favoured upstream and built-in forms of regulation. They proposed broader coverage across harmful content environments, system-level time and attention protections resistant to easy override, more consistent moderation and safer defaults, tighter regulation of advertising, and gentler interruption mechanisms that could create a pause or “return to reality” without becoming moralising or punitive. Education was valued as supportive, but not sufficient unless backed by governance levers capable of translating protective intent into consistent delivery. Finally, participants articulated a layered but asymmetrical model of responsibility. Responsibility was shared in principle, but not equally in practice. Platforms and government were positioned as central because they control, or can constrain, recommendation systems, content delivery, moderation infrastructures, reporting processes, commercial incentives, and enforceable standards. Parents and schools were recognised as supportive but bounded contributors rather than the primary site of risk management. Conclusion Taken together, the findings suggest that young people’s contributions are best understood not as simple policy preferences, but as sophisticated evaluations of mechanisms, trade-offs, and governance constraints. This thesis argues that youth perspectives should be treated as substantive governance insight for designing social media regulation that is protective, proportionate, enforceable, and compatible with continued digital participation. If harms are produced through structural conditions, credible protection must ultimately be pursued as structural change.