Australia banned under-16s from social media. Measuring whether it worked is the hard part

On 10 December 2025 the Australian government implemented Social Media Minimum Age legislation prohibiting Australians under 16 from holding social media accounts [s1]. Whatever else it is, it is an unusually clean natural experiment: a population-wide change to adolescent social media access, imposed on a known date, in a country with the research infrastructure to study it.

Three papers published across 2026 set out how that study is being attempted, what it can plausibly answer, and one specific way the law will make adolescent research harder to do at all.

The cohort with a pre-ban baseline

The design that comes closest to a controlled test is a four-wave longitudinal study whose protocol was published in BMJ Open on 5 August [s1]. Its central asset is timing: 422 young people aged 10 to 15, paired with a parent or carer, were recruited and completed a baseline assessment between October and December 2025 — before the legislation took effect [s1]. Follow-ups are scheduled at six months (April–May 2026), twelve months (October–November 2026) and eighteen months (April–May 2027) [s1].

A pre-intervention baseline is the thing almost every existing social media and mental health study lacks. Most of the literature compares heavier and lighter users at a single point in time, which cannot separate social media causing distress from distressed adolescents using more social media. Measuring the same children before and after an externally imposed access change removes some of that ambiguity, though not all of it.

The data collection is unusually granular for this field. Alongside adolescent and parent surveys, the protocol includes ecological momentary assessments, cognitive tasks, and smartphone-recorded screenshots of social media and messaging app use at each timepoint [s1]. Screenshots matter because self-reported screen time is a notoriously poor proxy for actual use. Analysis will use linear regression for correlational relationships, and linear mixed-effects models plus mixed-effects segmented regression to relate momentary experiences to more stable individual, parenting and governmental factors [s1]. The study was approved by the UNSW Human Research Ethics Committee under iRECS 9197 [s1].

This is a protocol. It reports no results. Nothing published so far tells anyone whether the law has changed adolescent mental health, and the first follow-up wave — April to May 2026 — has only recently closed. Any claim circulating that the ban has already been shown to help or harm is not coming from this study.

What a good evaluation would have to look at

A viewpoint published in The Lancet Regional Health – Western Pacific in February argues that the evaluation question is broader than mental health scores [s2]. It contends that comprehensive evaluation should examine second- and third-order effects: changes in young people's academic performance, digital literacy, mental and physical health indicators, and participation in alternative activities [s2].

That framing anticipates an obvious failure mode. If under-16s stop using regulated platforms and move to unregulated ones, or lose access to peer support they relied on, a study measuring only depression scores on the treated population could report a null result while missing the actual consequences.

The same viewpoint makes a structural argument: drawing on the experience of similar policies in Asian regions, it holds that age restrictions alone may achieve limited success without accompanying platform regulations built on safety-by-design principles [s2]. Specifically, it argues restrictions may be less effective without rules enforcing algorithmic transparency, duty-of-care obligations, and limits on user profiling [s2]. Its conclusion is that Australia should treat the policy as a starting point rather than an endpoint, establish rigorous long-term evaluation frameworks, and consider regulatory approaches addressing the structural incentives behind harmful platform design [s2].

The self-inflicted measurement problem

A second viewpoint in the same journal, published in March, identifies a complication that is close to circular [s3]. Social media has become a popular recruitment tool for adolescent research precisely because it enables rapid, low-cost, targeted enrolment [s3]. A mandatory minimum age of 16 is likely to disrupt researchers' ability to recruit adolescents aged 13 to 15, who are commonly reached through those platforms [s3].

The paper identifies three areas of impact: shifts in sample composition from reduced reach; increased reliance on gatekeeper-mediated recruitment, with risks to participation privacy and sample diversity; and greater operational complexity, longer timelines and higher costs [s3]. It notes flow-on effects for international studies that recruited Australian participants [s3].

The bind is worth stating plainly. The population whose wellbeing the law is meant to protect is now harder to enrol in the research that would establish whether it worked — and harder to enrol selectively, since the adolescents reachable through parents and schools are not a random sample of the adolescents who were reachable through Instagram. The paper's proposed remedies are structural: a national youth research registry, expanded funding and timelines to support multi-modal recruitment, and ethical, developmentally appropriate consent pathways that minimise gatekeeping [s3].

What to watch

The first genuinely informative datapoint will be results from the six- and twelve-month waves of the UNSW cohort, the latter not due until late 2026 [s1]. Even then, a 422-family single-cohort study is a starting point rather than a verdict, and it will need to be read alongside whether sample composition shifted in the way the recruitment paper predicts [s3].

Australia is running the experiment on a national scale either way. Whether it also runs the measurement well enough to learn from it is a separate decision, and one that has not been settled.

Sources

• [s1] Tracking mental health and cognitive changes following the Australian social media minimum age legislation: a 4-wave longitudinal study protocol, BMJ Open, published online 5 August 2026. https://doi.org/10.1136/bmjopen-2026-120216
• [s2] Beyond the debate: toward pragmatic evaluation of Australia's social media age restrictions, The Lancet Regional Health – Western Pacific, published online 21 February 2026. https://doi.org/10.1016/j.lanwpc.2026.101818
• [s3] How Australia's social media minimum age law may reshape research recruitment for adolescents under 16, The Lancet Regional Health – Western Pacific, published online 5 March 2026. https://doi.org/10.1016/j.lanwpc.2026.101822