Teens UnEdited: Claim Your Identity
the public research repository

The Evidence

Everything behind Teens UnEdited, in the open.

Every number below is drawn from a named primary source, dated to the survey wave it actually came from, and verified against the original before it appears here. This is the evidence, not the pitch, for the parents, educators and researchers who want to check the work.

5 rooms
of pressure
Primary
sources only, each named and dated
Verified
against the original
Contrary
evidence published too
Last verified September 10, 2026
Illustration of three teens with backpacks, the students behind the research on identity and belonging
what the data shows

One teenager, five rooms of pressure.

A teenager does not edit herself down for one reason. She does it because five concentric forces, each with its own logic, are all asking her to be a slightly different person. The innermost ring is who she actually is. Every ring outward adds another audience.

The research below is organized the way the pressure is actually organized, and read in order it shows both the shape of the problem and the single factor that shows up, in every ring, as the way out.

05 SCREENthe curated self04 PEERSthe pressure to fit03 SCHOOLwhen structure ends02 HOMEthe parent gap01 SELFwho they are

tap a ring to jump to its evidence

Watch: what Teens UnEdited is about

The five rooms, in under two minutes.

Self, home, school, peers, screens, and what AI changes about each of them. A 1:46 overview from Glenn Llopis Group.

01

Self

who they are

The innermost ring: identity, authenticity, and what it costs a young person to perform a version of themselves. This is the ring every other ring is pressing on.

33%

of U.S. high school students reported persistent feelings of sadness or hopelessness in 2025, down from 40% in 2023.

CDC Youth Risk Behavior Survey, 2025 (released September 2026) ↗
32%

of adolescents aged 12–17 received mental health treatment in the past year, roughly one in three.

SAMHSA, National Survey on Drug Use and Health: 2023 ↗
39 studies

A scoping review of the peer-reviewed record is consistent: authenticity in adolescents is tied to wellbeing and self-esteem, while inauthenticity is tied to depressive symptoms.

Adolescent Research Review: scoping review, 2023 ↗
New · 2026
39% vs 22%

Gen Z adults who are no longer in school, versus Gen Z students, who say it is only a little or not at all true that their life has meaning.

Gallup: “Many Gen Zers Missing Sense of Meaning or Purpose in Life,” June 2026 (fielded December 2025; n=2,436, ages 13–28) ↗
what helps
The UnEdited Index.

The measurable cost of inauthenticity is only actionable if someone can see it happening. The check-in gives a teenager language for what they are already editing, before the edit hardens into a habit they will spend their thirties undoing.

Take the 3-minute check-in →
02

Home

the parent gap

The first ring outward. The worry at home is real and it is high, but the channel it needs is narrower than parents think it is.

55%

of parents are extremely or very worried about the mental health of today's teens.

The concern is real, and higher than the teens' own.

Pew Research Center, 2025 (fielded Sept–Oct 2024) ↗
69 vs 44%

Parents versus teens who say it is harder to be a teenager today. Parents' top reason: social media.

Two generations see the same storm from different windows.

Pew Research Center, 2024 ↗
Half vs 80%

About half of teens are very comfortable talking with a parent about their mental health, while 80% of parents say they're comfortable having that conversation.

Parents feel ready to talk. Half of teens don't, yet. The gap is quiet, and it is the whole problem.

Pew Research Center, 2025 (fielded Sept–Oct 2024) ↗
what helps
The parent check-in, run alongside the teen's.

The gap here is not a shortage of care; it is a shortage of opening lines. Two parallel instruments give a parent and a teenager the same vocabulary and something specific to talk about, which is what turns 55% worried into one useful conversation.

Take the parent check-in →
03

School

when structure ends

The ring that holds a teenager up, and the cliff at its edge. This is also where the strongest evidence for the way out sits, because engagement and belonging are measurable here in a way they are nowhere else.

56%→39%

56% of Gen Z middle and high schoolers are thriving. Among Gen Z young adults, 39% are.

It falls exactly where school structure ends and self-leadership has to begin.

Gallup & Walton Family Foundation, Voices of Gen Z, 2025 wave (n=3,793, ages 13–28). The 2026 wave did not republish this comparison; this remains the current figure. ↗
Belonging spine
81 vs 27%

Highly engaged students who feel prepared for the future, versus the least engaged: a threefold gap.

Gallup counts twelve engaging school experiences, including interesting coursework and close connections to teachers and peers; “highly engaged” means ten or more, “least engaged” three or fewer. This is the crisis and the answer in one line.

Gallup & Walton Family Foundation, Voices of Gen Z, 2026 wave, published August 2026 (n=1,758 students aged 14–19; fielded May–June 2026) ↗
Belonging spine
Every measure

Feeling connected at school is linked to lower risk on every mental health measure the survey tracks.

Belonging is not one protective factor among many. It is the one that keeps showing up.

CDC, Youth Risk Behavior Survey, 2023 ↗
New · 2026
51%

of students agree that school gives them the opportunity to do what they do best, up from 37% in 2023.

Real improvement, and still barely half. The other half spend six hours a day somewhere that does not ask for their best thing.

Gallup & Walton Family Foundation, Voices of Gen Z, 2026 wave. Item: “At school, I get to do what I do best every day.” 37% (2023) → 43% → 50% → 51% (2026) ↗
New · 2026
29%

of students do not agree that there is a teacher or adult at school who regularly checks in on them to see how they are doing.

Nearly one in three has no reliable adult inside the building, which is the mechanism belonging runs on.

Gallup & Walton Family Foundation, Voices of Gen Z, 2026 wave (71% agree; 29% do not) ↗
New · 2026
45%

of students do not agree that their school is the perfect place for people like them.

Most of that is hesitation rather than rejection (31% land on “neither”), but a teenager who is unsure whether they fit is a teenager deciding what to edit.

Gallup & Walton Family Foundation, Voices of Gen Z, 2026 wave (55% agree; 14% actively disagree) ↗
what helps
The curriculum, inside the school day.

The 81%-versus-27% gap is the whole argument for doing identity work where the structure still exists. Engagement is not a mood; Gallup counts it as twelve concrete experiences, most of which a school can actually build. Belonging is buildable, and this is the ring where you build it.

See the curriculum →
Belonging is not one protective factor among many. It is the one that keeps showing up.
04

Peers

the pressure to fit

The ring where the template arrives. Not just who to be, but which path to take, and, for the eight million teenagers who play a sport, an identity with a market price attached.

88%

of teens feel pressure to follow a specific educational or career path.

The template arrives before the identity does.

Junior Achievement USA & Citizens, Teens & the Future of Work: 2025 (n=1,000, ages 13–18; fielded July–August 2025) ↗
41%

of teens feel pressure to fit in socially: 45% of girls and 37% of boys.

The pressure is not evenly distributed, and it lands hardest on the group already reporting the worst wellbeing.

Pew Research Center, The Gender Gap in Teen Experiences: March 2025 ↗
31%

say what they see on social media makes them feel like their friends are leaving them out: 36% of girls, 26% of boys.

Exclusion used to end when you left the building. Now it is visible all evening.

Pew Research Center, Social Media and Teens' Mental Health: April 2025 ↗
8.26M

high school athletes in the United States, an all-time record. About 560,000 of them, roughly 7%, will compete at an NCAA school.

For the rest, an identity built entirely on the sport has a fixed expiry date, and almost nobody is helping them plan for it.

NFHS, High School Athletics Participation Survey: 2024–25 (8,260,891 participants) ↗NCAA: Probability of Competing Beyond High School, 2024–25. NCAA notes these ratios are approximations. ↗
what helps
The books, and Athlete UnEdited for the NIL cohort.

A teenager with a claimed identity has something to hold when the room pushes. For the athlete the stakes are sharper and the timeline is shorter: NIL makes the performed self financially rational, which is exactly when someone needs to be asked who they are when the sport stops.

Athlete UnEdited →
05

Screen

the curated self

The outermost ring, and now two things at once: the social platforms that ask a teenager to perform a self, and the AI that will agree with whatever self they perform.

48%

of teens say social media has a mostly negative effect on people their age, up from 32% in 2022.

Performing a version of yourself is now the majority experience. See what cuts against us: teens rate their own experience far more positively than they rate it for their peers.

Pew Research Center, Social Media and Teens' Mental Health: April 2025 (fielded Sept–Oct 2024) ↗
72%

of teens have used an AI companion, a digital character built for conversation that feels personal.

Among those users, 33% have chosen to discuss important or serious matters with an AI companion instead of a real person.

Common Sense Media, Talk, Trust, and Trade-Offs: July 2025 (n=1,060, ages 13–17). The most recent companion-specific measure; the 2026 census below measures something broader. ↗
New · 2026
86%

of children aged 9–17 now use or interact with AI, including about one in four who do so daily.

Among those who have discussed their feelings or personal problems with AI, 25% say it sometimes understands them better than most people do.

Common Sense Media Census: AI Use by Tweens and Teens: June 2026 (n=1,204, ages 9–17; fielded March 2026). Broader measure than the 72% above; the two are not a trend line. ↗
New · 2026
19.2%

of 12–21-year-olds have used an AI chatbot for advice when feeling sad, angry, nervous or stressed. Of those, 63% told no one they had done it.

The turn to the machine is already happening privately, at scale, and outside every system built to notice.

JAMA Pediatrics: “AI Chatbot Use and Disclosure for Mental Health Among US Adolescents and Young Adults,” June 3, 2026 (RAND-led; n=1,009; fielded November 2025) ↗
67%

of students say using AI for schoolwork is harming their own critical thinking, up from 54% in a single year, while use rose from 48% to 62%.

They can feel it happening.

RAND American Youth Panel: December 2025 ↗
31%↑

Anger at AI among Gen Z, up 9 points in a year, while excitement fell from 36% to 22%. And 74% now say AI is likely to make learning more difficult for them.

This is the turn: they are not sleepwalking into it. They are looking for the alternative.

Walton Family Foundation, GSV Ventures & Gallup, The AI Paradox: 2026 ↗
what helps
The identity work the platforms cannot do.

A teenager who has never been asked who they are will accept the version an algorithm reflects back, and an AI that agrees with everything is the most efficient identity-editor ever built. The 31% anger figure matters most here: this generation is already looking for the alternative. The work is to have one ready.

See the books →
kept in view

What cuts against us.

A research repository that only publishes its own supporting evidence is a brochure. These are the strongest findings we know of that complicate, weaken or contradict the case above. They are here because they are true, and because anyone serious about this will find them anyway.

20.8%→15.1%

Adolescent major depressive episodes fell from 20.8% in 2021 to 15.1% in 2025, a statistically significant decline in the federal government's own flagship survey. Serious thoughts of suicide fell from 12.9% to 10.4% over the same period.

This is the single strongest piece of evidence that the acute crisis may be easing.

SAMHSA, National Survey on Drug Use and Health: 2025, published July 2026 (10,933 adolescent interviews) ↗
39.1%→29.7%

North Carolina's own high school students, 2023 to 2025: persistent sadness or hopelessness down 9.4 points in two years, within the same state and the same instrument.

State-level 2025 results are arriving before the national wave, and several point the same way.

North Carolina Youth Risk Behavior Survey: 2025, released August 2026 (n=1,692) ↗
23%→18%

Severe depression among college students fell for a third consecutive year. Moderate-to-severe anxiety fell from 37% to 32%, and those seriously considering suicide from 15% to 11%.

The cohort just past the cliff is improving, not worsening.

Healthy Minds Study: 2024–25, published September 2025 (n>84,000 across 135 institutions) ↗
~60%

of teens report their own experience on TikTok, Instagram and Snapchat is mostly positive; only 3% call it mostly negative.

Set against the 48% who say social media is bad for people their age, this is the classic self-versus-others gap, and it means the 48% figure measures a perception, not a lived experience.

Pew Research Center, Teens' Experiences on TikTok, Instagram and Snapchat: April 2026 ↗
85%+

of under-16s in an Australian cohort were still using restricted platforms three months after the world's first national ban took effect. The regulator's own larger evaluation found account ownership fell only from 52% to 42%, with over half of children never asked to prove their age.

Restriction alone does not appear to work.

The BMJ: Barnes et al., June 2026 (n=408) ↗eSafety Commissioner: three-month evaluation, July 2026 (n>4,000) ↗
Ch. 4

The World Happiness Report 2026 devoted its third chapter to the case that social media is harming adolescents at population scale, and its fourth to the rebuttal.

We cite Chapter 3 because it is rigorous and relevant. We note Chapter 4 because the editors deliberately published both, and the field is not settled.

World Happiness Report 2026: Ch. 3, Haidt & Rausch; Ch. 4, response. Published March 2026. ↗

Where that leaves the case

Two things are true at once. The acute clinical measures, depression and suicidal ideation, are improving from their 2021 peak, and we should say so first and loudest. And the identity measures are not: the thriving cliff has no verified successor showing it has closed, nearly one in three students still has no adult who checks in on them, and AI use has moved from novelty to daily infrastructure in eighteen months.

Our thesis was never that teenagers are in acute crisis. It is that they are quietly editing themselves down, and that the editing is not what a depression screener measures. The contrary evidence sharpens that distinction rather than dissolving it, but if the identity measures follow the clinical ones downward, we will publish that here too.

the institutional response

The world is catching up to what parents already know.

In five years the response has escalated from advisories to laws to liability, and now to moral authority and money. The concern felt at the kitchen table is now the position of surgeons general, parliaments, regulators, juries, and the Vatican.

Dec 2021

The U.S. Surgeon General declares a youth mental health crisis

Protecting Youth Mental Health names it a national emergency, not a family-by-family problem.

U.S. Surgeon General advisory ↗
Jun 2022

Congress puts $1 billion behind school mental health staffing

The Bipartisan Safer Communities Act funds the two federal grant programs for school-based mental health professionals. The follow-through has been uneven: after the Department of Education cancelled up to $1B in school mental health grants in 2025, it announced $208M in new, differently-conditioned grants to 65 recipients in December 2025, roughly a fifth of what was cancelled, and restricted to school psychologists.

U.S. Department of Education: BSCA mental health grants ↗
May 2023

A second advisory targets social media itself

Social Media and Youth Mental Health puts the platform environment, not the teenager, under scrutiny.

U.S. Surgeon General advisory ↗
Jun 2024

The Surgeon General calls for tobacco-style warning labels

Writing in The New York Times, the nation's top doctor frames the platforms as a product requiring a health warning.

Mar 2025

Utah becomes the first state to make app stores verify ages

The App Store Accountability Act (S.B. 142) shifts the burden of age assurance to the platforms that distribute the apps. Its core obligations took effect in May 2026, with enforcement following at the end of that year.

Utah S.B. 142 (enrolled) ↗
Sep 2025

The FTC opens an inquiry into AI companions and child safety

Unanimous 6(b) orders to seven companies (Alphabet, Character Technologies, Instagram, Meta, OpenAI, Snap and X.AI) demanding to know what they have done to evaluate the safety of chatbots acting as companions to children and teens.

U.S. Federal Trade Commission ↗
Oct 2025

California passes the first U.S. law regulating AI companions

S.B. 243 imposes disclosure duties, crisis and self-harm protocols, and minor-specific protections on companion-chatbot operators. In force since January 2026.

California S.B. 243 bill text ↗
Dec 2025

Australia's under-16 social media ban takes effect

The first national minimum-age law in the world, covering ten platforms. Early evaluations show widespread circumvention: see what cuts against us.

eSafety Commissioner ↗
Mar 2026

A California jury finds Meta and YouTube injured a young user by design

In Los Angeles Superior Court, the first verdict of its kind: both companies found negligent in platform design and liable for failure to warn, with $3M compensatory and $3M punitive damages, and a finding of malice, oppression or fraud. Both are appealing. More than 2,400 similar federal actions remain pending.

May 2026

Pope Leo XIV devotes his first encyclical to artificial intelligence

Magnifica Humanitas insists that school remain “the place where new generations can learn to seek and love the truth, to reflect on the meaning of life and to recognize the dignity of every person,” and warns that the desire to ask questions must not be extinguished in the young by machines that make human thought seem useless.

Vatican News: Magnifica Humanitas ↗
Jun–Aug 2026

Children's online safety legislation moves in both chambers

The House passed the KIDS Act 267–117 in June, folding in KOSA, COPPA 2.0 and AI-chatbot requirements; KOSA itself cleared Senate Commerce in August with 76 cosponsors. Neither is law yet. Illinois signed its own Children's Social Media Safety Act in July, requiring chronological feeds for minors and overnight notification blackouts.

Aug 2026

Meta settles with the states for up to $17 billion

Two weeks into the state attorneys general trial in Oakland, Meta settled: approximately $12.2 billion guaranteed over ten years, rising to a maximum near $17 billion if other major platforms adopt comparable time-management and age-assurance measures. No admission of liability. The settlement expressly preserves the personal-injury and school-district cases, which continue.

Office of the California Attorney General ↗
the economic picture

Even the market is tracking this generation.

Beyond the health data, the institutions that study consumers and capital are documenting the same shift: a generation under unusual pressure, searching for something real, and increasingly priced as a liability by the platforms built to hold their attention.

Market-sizing figures below come from commercial research houses and are labeled as such; they are directional evidence of commercial interest, not measurements.
$17B

the size of Meta's August 2026 settlement with the states over youth addiction claims: roughly $12.2B guaranteed, the balance contingent on industry-wide adoption of time limits and age assurance.

Platform design is now being priced in the language once used for tobacco.

California Attorney General: consent judgment, August 26, 2026 ↗
Commercial estimate
$4.3–4.7B

the estimated size of the global social-emotional learning market in 2025, with projected annual growth ranging from about 19% to 26% depending on the firm.

The category exists and is funded; what it mostly lacks is identity work.

Grand View Research ($4.3B, 26.3% CAGR) ↗IMARC Group ($4.72B, 18.6% CAGR). Commercial estimates, methodology not public. ↗
11 percentile pts

the academic achievement gain associated with school-based social-emotional learning, a result first established across 213 programs in 2011 and independently reproduced on modern studies in 2025.

The mechanism this page keeps pointing at is one schools can already be shown a return on.

Durlak et al., Child Development (2011; 213 programs, 270,034 students) ↗Ha, McCarthy, Strambler & Cipriano, Review of Educational Research (2025) ↗
40%

of Gen Z say they feel stressed or anxious all or most of the time, and just 52% rate their mental wellbeing as good. Yet 89% say a sense of purpose matters to their job satisfaction and wellbeing.

The demand is explicit.

Deloitte, Gen Z & Millennial Survey: 2025 edition (fielded late 2024; n=23,482 across 44 countries). Deloitte's 2026 edition reports mental health improving. ↗
NIL wedge
$4.5B

Opendorse's revised projection for the college NIL market in 2026–27, up from its own earlier $2.8B estimate. High school NIL is now permitted in all but five states.

An identity with a market price attached, arriving earlier every year.

Opendorse: July 2026 revision; state tracker updated April 2026 (Alabama, Hawaii, Indiana, Mississippi and Wyoming still prohibit). Vendor estimates from proprietary transaction data. ↗
$336

the core beauty “wallet” in Piper Sandler's semi-annual teen survey, down 2% year over year, with total self-reported teen spending down 6%. The record $374 set in spring 2025 has already turned.

Money is coming out of the outward, edited self.

Piper Sandler, Taking Stock With Teens: 50th wave, October 2025 (n=10,959). Now an annual survey. ↗
The through-line across every source: this generation values individual expression, distrusts labels, and is searching, in McKinsey's phrase, for what is true. That is the demand the evidence keeps pointing at.
Research

What the research says

Each of these essays starts with published work by leading scholars of adolescence, then asks what it means for your family and your teen. Every researcher is credited, and every study is linked.

Screen

Why Losing Likes Stings Teens More

Drawing on research by
da Silva Pinho et al. (2024); Hofmans & van den Bos (2025)
4 min read
Self

Let Them Wander: Exploration and a Teen's Good Days

Drawing on research by
Saragosa-Harris et al. (2022); Insel & Cohen (2025)
5 min read
School

Different People Hold Up Different Parts of Your Teen's School Life

Drawing on research by
Wentzel et al. (2016); Wentzel (1997)
5 min read
Screen

Five Days Without Screens, and What Kids Could Read in Faces

Drawing on research by
Uhls et al. (2014); Uhls et al. (2017)
5 min read
School

Belonging Programs May Be Built for the Outgoing Kid

Drawing on research by
Stubblebine et al. (2024); Brady & Gopalan (2025)
5 min read
School

Good Intentions Aren't Enough for School Wellbeing Programs

Drawing on research by
Guzman-Holst et al. (2025); Foulkes et al. (2024)
4 min read
how we verify

Every number here can be checked.

This repository has one rule: nothing appears until it has been traced to a named primary source and verified against the original publication. No secondary summaries, no rounded-up marketing figures, and no quietly relabeled dates.

1

Primary sources only

Every figure is attributed to the organization and study that produced it (the CDC, SAMHSA, Pew, Gallup, peer-reviewed journals), not to the news write-up about it.

2

Dated to its own wave

Each figure carries the year of the survey wave it actually came from. We do not relabel a 2025 figure as 2026 to look current, and we do not drop a strong figure because it is a year old.

3

Contrary evidence published

Findings that weaken or contradict our case get their own section, with the same detail as the findings that support it. If the data turns, this page turns with it.

4

A living record

Last verified September 10, 2026. New research is reviewed as it is published and mapped against existing claims to strengthen, update or challenge them.

Full source list

Dated to the survey wave or publication each figure came from. Each links to the original publication.

Centers for Disease Control and Prevention: Youth Risk Behavior Survey, 2023 (no national 2025 wave published as of September 2026) ↗SAMHSA: National Survey on Drug Use and Health, 2023 ↗SAMHSA: National Survey on Drug Use and Health, 2025 ↗North Carolina Department of Public Instruction / NC DHHS: NC Youth Risk Behavior Survey, 2025 ↗Gallup & the Walton Family Foundation: Voices of Gen Z, 2025 wave ↗Gallup & the Walton Family Foundation: Voices of Gen Z, 2026 wave ↗Gallup: Gen Z meaning and purpose, June 2026 (fielded December 2025) ↗Walton Family Foundation, GSV Ventures & Gallup: The AI Paradox, 2026 ↗Pew Research Center: teen and parent survey, 2024 ↗Pew Research Center: The Gender Gap in Teen Experiences, March 2025 ↗Pew Research Center: Teens, Social Media and Mental Health, April 2025 ↗Pew Research Center: Teens' Experiences on TikTok, Instagram and Snapchat, April 2026 ↗Common Sense Media: Talk, Trust, and Trade-Offs, July 2025 ↗Common Sense Media: AI Census, June 2026 ↗JAMA Pediatrics: AI chatbot use and disclosure for mental health, June 2026 ↗RAND: American Youth Panel, December 2025 ↗Junior Achievement USA & Citizens: Teens & the Future of Work, 2025 ↗Adolescent Research Review: authenticity scoping review, 2023 ↗
the research behind it

Teens aren't broken. Many are editing themselves to fit.

Drawn from the work of 92 academic researchers across four fields: brain development, identity and belonging, social media and AI, and school. The headline is a conclusion across the studies, not a single measured finding. Social media research is contested, so both sides are included.

US high schoolers persistently sad or hopeless

30% in 2015, 42% in 2021, 40% in 2023, 33% in 2025.

Improving, still above 2015. CDC Youth Risk Behavior Survey, 2025 results released September 2026.

Things are getting a little better. Not better enough. And a new kind of companion just arrived during the years teens are working out who they are.

1The mid-teens are a sensitive windowSensitivity to peers peaks early. Self-control keeps maturing into the early 20s.
  • Matures early: self-consciousness and sensitivity to how peers judge you; reward from social approval; rebuilding a sense of self from how others see you.
  • Keeps maturing into the early 20s: self-control and planning; bouncing back after rejection; calming stress on your own. By the mid-teens, a parent's presence no longer buffers stress the way it did.
  • The gap between the two is a window of both risk and opportunity.

Blakemore (Cambridge); Steinberg (Temple); Luna (Pittsburgh); Somerville; Casey; Galván; Pfeifer; Dahl; Tottenham; Silvers.

Read: "The Teen Brain Isn't Done Until 25" Is a Script, Not a Finding →

2Teens adapt how they present themselvesAcross school, feeds and friend groups, studies find teens editing to fit.
  • Performing school: high achievers learn to perform school rather than engage with it. Pope, Stanford
  • Holding back online: posts are public, permanent and seen across contexts, so teens self-censor. Weinstein, Harvard
  • Hiding parts of themselves: teens can name which parts a setting won't accept, and edit accordingly. Rogers; Covarrubias, UC Santa Cruz
  • Acting older for status: 13-year-olds who did it had more problems in their 20s. J. Allen, Virginia
  • Keeping secrets: concealment wears down a felt sense of authenticity over time. Slepian, Columbia

Read: Your Teen Isn't Just Scrolling. They're Managing an Audience. →

3Social media's effects differ teen to teenThe average effect is small because some teens benefit and others are harmed.
46%
felt better after passive use
44%
felt no different
10%
felt worse
  • Alarm: phone-based childhood drove rising anxiety and depression. Haidt; Twenge
  • Skeptic: average effects are small and haven't grown. Odgers; Orben; Przybylski
  • Differential: content and comparison matter more than minutes; some teens find belonging online. Valkenburg; Burnell; Charmaraman
  • 48% of teens say social media is mostly negative for people their age. Only 14% say that about themselves.

Beyens, Valkenburg et al., Scientific Reports, 2020 (2,155 real-time check-ins); Pew Research Center, April 2025.

Read: Social Media Isn't Good or Bad for Your Teen. It's Personal. →

4AI companions are an emerging concernMost teens use chatbots now. The research is new, and use is rising.
64%
of US teens use AI chatbots
3 in 10
use them every day
1 in 3+
young AI users have talked through feelings or problems with AI
  • An empathy gap: chatbots simulate care without the healthy conflict through which teens learn closeness and boundaries. Children are likely to miss the difference. Kurian, Warwick; Vasan, Stanford
  • A loneliness link: kids who use AI daily or weekly are more likely to report feeling lonely, and a quarter say AI understands them better than most people. Common Sense Media, 2026
  • Safety gaps: researchers posing as teens found AI companions readily produced sexual, self-harm and violent content. Vasan, Stanford
  • The American Psychological Association issued a 2025 health advisory on AI and adolescents.

Pew Research Center, How Teens Use and View AI, February 2026; Common Sense Media, June 2026.

Read: A Chatbot Can Keep Teens Talking →

5Belonging is a strong protective factorBrief, well-timed belonging work shows effects lasting years.
30–37%
lower rates on every mental-health and suicide-risk measure for students highly connected to school
1 hour
belonging exercise with effects lasting years
  • Belonging is most easily lost at the move into high school. Benner, UT Austin
  • Settings work when teens are known as whole people. Deutsch. Exploring identity builds resilience. Del Toro; Umaña-Taylor

CDC, MMWR Supplement, YRBS 2023; Walton and Cohen, Science, 2011.

Read: Belonging Isn't Fitting In →

6What adults do shapes outcomesHigh standards with high support beat monitoring and control.
  • Mentor mindset: adults who pair high standards with support earn the trust that lets teens take feedback. Yeager, UT Austin
  • Meaning over monitoring: involvement that links school to a future raises achievement; surveillance doesn't. Hill, Harvard
  • Expectations matter: teens who expect adolescence to be stormy do worse, and that belief can change. Qu, Northwestern
  • Well-run social-emotional learning raises academic results by about 11 percentile points. Durlak, 213 studies
  • A caution: awareness campaigns may lead some teens to read ordinary distress as disorder. Foulkes, Oxford

What the evidence points to

Build belonging · change how adults show up · reach teens early, especially around the move into high school · make it safe not to self-edit · prepare for AI companions · ask "for whom?" before setting social media rules.

keep reading

What the research says, in plain words.

Each one starts from a published study and asks what it means at home and at school.

More research reads →
what to do with it

Five rooms, one lever: belonging.

The research names the pressure in every ring, and the same protective factor keeps appearing inside all five. An adult has to reclaim a self that was eroded. A teenager only has to claim one that is still forming.

optional

Send me the research

This page stays open to everyone. If you'd like the research in your inbox to share with your family, your school or your board, leave your name and email. We'll send it right away, plus one short follow-up a couple of days later.

Thank you. The research is on its way to your inbox.
Something went wrong. Please check your email address and try again.