TikTok and Teenage Mental Health: Both Camps Are Overselling Their Certainty
What the evidence actually supports once you strip out the moral panic and the corporate spin — and the one-hour rule we landed on at home.
by Gary Worthington, More Than Monkeys
I sat down a few weeks ago to write my fourteen-year-old daughter a straight, factual case for why TikTok is capped at an hour a day in our house. Twenty years of reading systems for a living — work out what a thing was actually built to do, not what the documentation claims, and the rest follows — made me assume this would be a quick job.
It wasn’t.
What I found instead was two well-populated camps, both talking with far more confidence than the underlying evidence earns. One camp treats “the algorithm is rewiring children’s brains” as settled fact. The other treats “it’s just correlation, calm down” as the final word. I went looking for what would actually survive a cross-examination from a sharp teenager, and ended up somewhere less comfortable than either side: the evidence is messier, smaller in places, and stronger in places, than either camp wants to admit.
The panic and the pushback
Jonathan Haidt’s The Anxious Generation made the moral-panic case loudest: phones and social media rewired an entire generation, full stop. It sold enormously and shaped a real legislative agenda. It also drew a serious academic rebuttal — Candice Odgers, reviewing the same territory in Nature, points out that hundreds of researchers going looking for Haidt’s effect sizes have mostly come back with null, small, or mixed results, and that most of the underlying data is correlational rather than causal.
Both of those things can be true at once, which is the bit that gets lost. A weak-to-moderate correlation across dozens of studies is not nothing, and “it’s mostly correlational” is not the same claim as “there’s no effect.” Reading the actual papers rather than the press releases either side puts out, the honest position is: there’s a real, specific, partially-evidenced set of harms here, sitting under a lot of noise generated by people who’ve already decided how the story ends.
That’s the position I ended up defending to my daughter. Not “TikTok is destroying you.” Not “there’s nothing to see here.” Just: here’s what’s actually been measured, here’s how solid each bit is, and here’s why I still think an hour a day is the right number.
What the company’s own paperwork says
The one part of this that isn’t contested is intent. You don’t need a psychologist for this bit — you need the discovery documents from a lawsuit.
Litigation against TikTok has forced the release of internal material that reads less like a tech company’s PR and more like an admission. One document states plainly that using the app creates “a strong and constant urge to keep opening the app.” A TikTok executive is on record calling teenagers “the golden audience” — the cohort you most want hooked, because they’re the most valuable to keep engaged. CNN’s reporting on the unsealed litigation filings quotes a similar internal note from a competitor’s own researchers: “IG is a drug… we’re basically pushers.”
European regulators reached the same conclusion from the outside rather than the inside. The European Commission’s preliminary findings under the Digital Services Act state that TikTok’s endless feed, autoplay, and constantly re-tuned recommendation engine are built specifically to foster compulsive use rather than merely to serve relevant content.
If I shipped a system that measurably created “a strong and constant urge” to keep using it, and my own internal docs said so, nobody at my next client engagement would call that an accident. They’d call it the spec.
None of this required TikTok to invent anything new. Nir Eyal published the playbook back in 2014, in a bestseller called Hooked: How to Build Habit-Forming Products, written explicitly for product teams and growth marketers. His Hook Model runs on a four-stage loop — trigger, action, variable reward, investment — and it maps onto TikTok with almost embarrassing precision: the notification or the boredom is the trigger, one swipe is the action stripped of all friction, an unpredictable feed of clips is the variable reward (Eyal uses slot machines as the explicit comparison in the book, not a critic’s), and following, liking, and building a watch history the algorithm learns from is the investment that primes the next loop. Eyal himself went on to write a second book, Indistractable, about escaping exactly this pull — five years after teaching the industry how to build it. His own Wikipedia page notes that critics have compared his personal-responsibility framing to the tobacco industry’s old line: keep the focus on individual willpower, and corporate accountability quietly leaves the room.
I’m not writing this from a position of innocence, either. We used the Hook Model directly when I was building Yepic — trigger, action, variable reward, investment, laid out on a whiteboard on purpose, because that’s what makes a product stick. It works, and that’s exactly the problem.
In January this year, TikTok settled the first major US bellwether trial over exactly this claim rather than let a jury see the underlying evidence. A settlement isn’t a verdict — companies settle for all sorts of reasons that have nothing to do with guilt. But it tells you which way the company’s own lawyers thought the wind was blowing.
Sleep is the one nobody can seriously argue with
Everything above is about intent. This next bit is about outcome, and it’s the least contestable finding in the whole area: screen use late at night delays and degrades sleep, and poor sleep reliably worsens anxiety, low mood, and concentration.
No caveats needed here. This single mechanism is, on its own, a complete justification for keeping devices out of bedrooms overnight — which is a rule in our house regardless of anything else in this article.
Body image, and the number that doesn’t quite hold up the way it’s quoted
The most-cited statistic in this entire debate is the “one in three teen girls” figure from Meta’s leaked internal Instagram research — that the app made existing body image problems worse for roughly a third of teen girls surveyed.
It’s worth knowing where that number actually comes from, because most people repeating it don’t. NPR’s reporting on the same leaked documents notes that Facebook’s own annotations show the finding was drawn from about 150 respondents out of a few thousand surveyed — and only those who’d already reported body image issues were asked the follow-up question at all. It’s a subset of a subset of a subset, dressed up in a slide deck as a headline stat.
That doesn’t make the underlying concern fake. The same tranche of research found roughly half of teen girls regularly compare their appearance to what they see on the platform, drawn from a properly-sized 50,000-person, ten-country survey — a far sturdier number than the famous one-in-three. TikTok runs on the same short-form, appearance-heavy format, and its own leaked documents raise the same concerns. The honest version of this argument uses the sturdy number, not the viral one.
Anxiety, and the strangest piece of evidence in the whole area
A cross-sectional study of Greek adolescents found a direct association between problematic TikTok use and anxiety symptoms — heavier compulsive use, higher anxiety scores. Cross-sectional means exactly what it sounds like: a single snapshot, no ability to say which came first. Take it as one data point, not a verdict.
The stranger, better-evidenced finding sits in neurology journals rather than psychology ones. Starting in 2021, clinicians in several countries independently documented teenage girls — with no prior history of tic disorders — abruptly developing tic-like movements after watching TikTok creators describe identical symptoms. The original case series, published in Movement Disorders Clinical Practice, is careful and clinical about it, and the phenomenon now has a name in the medical literature: TikTok tics, a form of mass social-media-induced illness. Pre-existing anxiety made people more susceptible to picking the symptoms up after exposure.
Whatever else you make of that, it’s peer-reviewed evidence that psychological and even physical symptoms can propagate through this platform the same way a dance trend does. That’s not a metaphor for how the app works. It’s a demonstrated mechanism.
The correlation everyone quotes, and the experiment that actually matters more
The headline correlational finding, repeated in nearly every parenting article on this topic, is a dose-response meta-analysis finding a 13% increase in depression risk for each additional hour of daily social media use among adolescents, with a stronger effect in girls. It’s a real, well-conducted piece of research. It’s also correlational, and correlational data can’t rule out the obvious alternative explanation: teenagers who are already struggling reach for their phones more, not the other way round.
The evidence that actually moves the needle on causation is quieter and gets quoted far less. A 2025 meta-analysis pooling ten randomised controlled trials — real experiments, where people were assigned to cut back or carry on, rather than just observed — found that reducing social media use produced a measurable drop in depression symptoms, even after adjusting for publication bias.
That’s the whole argument, really. Not “heavy users are more depressed,” which tells you nothing about direction. “People who cut back got measurably better,” which does.
What I’m not going to claim
I’m not going to tell my daughter this is settled science, because it isn’t. Nobody has a clean causal chain from “opened TikTok” to “developed depression.” What exists is a company that documented, in its own words, an intent to create compulsive use; a sleep mechanism that isn’t in dispute at all; a body image effect that’s real but smaller and messier than its own famous headline number; a genuinely strange and well-documented case of psychiatric symptoms spreading through the platform; and experimental evidence — not just correlation — that cutting back improves mood.
Put together, that’s not a slam dunk. It’s also not nothing, and it’s considerably more than “there’s no evidence.”
What we’ve actually agreed at home
An hour a day, tracked through the phone’s own downtime tools, no negotiation on the number. No phone in the bedroom overnight — the sleep evidence alone earns that one outright. And she gets to read the actual sources rather than a parent’s paraphrase of a moral panic, because if the case is real, it survives her checking my working. If it isn’t, it shouldn’t survive at all.
Final thought
The two camps shouting at each other about this — one insisting the science is settled, the other insisting there’s nothing to see — have both skipped the step where you actually read the studies. The real picture is smaller than the panic and bigger than the shrug. An hour a day, phone out of the bedroom, and a teenager who knows exactly what she’s opening and why it was built that way. That’s the whole policy. It didn’t need a moral panic to get there, and it doesn’t need one to hold.
Gary Worthington is a software engineer, delivery consultant, and fractional CTO who helps teams move fast, learn faster, and scale when it matters. He writes about modern engineering, product thinking, and helping teams ship things that matter.
Through his consultancy, More Than Monkeys, Gary helps startups and scaleups improve how they build software — from tech strategy and agile delivery to product validation and team development.
Visit morethanmonkeys.co.uk to learn how we can help you build better, faster.
Follow Gary on LinkedIn for practical insights into engineering leadership, agile delivery, and team performance.
Sources
Odgers, C. “The great rewiring: is social media really behind an epidemic of teenage mental illness?” Nature, 29 March 2024. nature.com/articles/d41586–024–00902–2
Duffy, C. “Lawsuit alleges social media giants buried their own research on teen mental health harms.” CNN Business, 25 November 2025. cnn.com/2025/11/25/tech/social-media-youth-mental-health-lawsuit-meta-tiktok-snap-youtube
European Commission preliminary findings on TikTok, Digital Services Act, summarised by 5Rights Foundation, 2 March 2026. 5rightsfoundation.com/tiktoks-addictive-design-preliminarily-found-in-breach-of-the-digital-services-act
Eyal, N. Hooked: How to Build Habit-Forming Products. Portfolio/Penguin, 2014. Model summarised at mindtools.com/aapqtdb/the-hook-model-of-behavioral-design
“Nir Eyal.” Wikipedia, accessed July 2026. en.wikipedia.org/wiki/Nir_Eyal
“Meta and YouTube head to trial over harm to children after TikTok settles.” NPR, 27 January 2026. npr.org/2026/01/27/nx-s1–5684196/social-media-kids-addiction-mental-health-trial
Shapiro, A. “The impact of Facebook and Instagram on teens isn’t so clear.” NPR, 6 October 2021. npr.org/2021/10/06/1043138622/facebook-instagram-teens-mental-health
“Research on Instagram and teens: Summaries from the Facebook Files.” Fairplay, 28 October 2021. fairplayforkids.org/research-on-instagram-and-teens-summaries-from-the-facebook-files
“Association Between TikTok Use and Anxiety, Depression, and Sleepiness Among Adolescents: A Cross-Sectional Study in Greece.” PMC. pmc.ncbi.nlm.nih.gov/articles/PMC11932271
Hull, M., Parnes, M., Jankovic, J. “Tics and TikTok: Functional Tics Spread Through Social Media.” Movement Disorders Clinical Practice, 2021. movementdisorders.onlinelibrary.wiley.com/doi/10.1002/mdc3.13267
Liu, M. et al. “Time Spent on Social Media and Risk of Depression in Adolescents: A Dose-Response Meta-Analysis.” PubMed, 2022. pubmed.ncbi.nlm.nih.gov/35564559
“Reducing Social Media Use Decreases Depression Symptoms: A Meta-Analysis of Randomised Controlled Trials.” MDPI, 2025. mdpi.com/2254–9625/15/11/222
Tags: TikTok, Parenting, Social Media, Mental Health, Technology Ethics