In March 2026, three psychiatric epidemiologists published a paper in JAMA. NPR covered it. The headlines noted, almost with surprise, that screens might not be to blame. Then everyone went back to telling teenagers to put their phones down.
The researchers had pulled the largest available national dataset on adolescent sleep, the Youth Risk Behavior Survey, and traced what had happened to American teenagers’ sleep between 2007 and 2023.
The finding was that adolescent sleep deprivation had increased over those years. The finding that mattered was that the increase showed up across every behavioral risk category the researchers tested.
The kids who used social media heavily slept less. So did the kids who did not.
The kids who drank alcohol slept less. So did the kids who did not.
The kids who played video games into the night slept less. So did the kids who never touched a console.
The kids whose parents had no rules about bedtime slept less. So did the kids whose parents enforced bedtimes rigorously.
The trend was systemic. It was not a behavior problem at the individual-teenager level. It was something happening to the entire developmental cohort across categories that should have predicted very different outcomes. This is what an environmental signal looks like in epidemiology. When a phenomenon shows up across populations that should have insulated against it, the cause is upstream of the populations.
The Bommersbach finding sits on top of more than two decades of work by Ronald Dahl and other developmental sleep scientists, who established in the early 2000s that pubertal neurobiology shifts the adolescent circadian rhythm later by about two hours. Teenagers are not lazy. They are biologically incapable of falling asleep at 10 PM and waking at 6:45 AM with anything like a full night’s sleep behind them.
We built a school system that requires them to do exactly that. Then we built the habit of blaming them for the consequences.
The Bommersbach Data
The 2026 JAMA paper by Tanner Bommersbach, Mark Olfson, and Taeho Greg Rhee did one specific thing very well. It pulled the trend in adolescent sleep duration alongside fifteen different behavioral risk markers and asked whether the trend in sleep loss was concentrated in any specific subgroup.
It was not. Insufficient sleep among US high school students rose from 69% in 2007 to 77% in 2023, with the steepest rise in very short sleep (under five hours per night). When the researchers split the data by behavioral risk factor, the rate of sleep loss increased in roughly equal measure across every behavioral group they tested.
The dominant explanation for adolescent sleep loss has been individual behavior — phones, social media, late-night gaming, lax parenting, a generational character flaw. The Bommersbach data say the individual-behavior explanation is at minimum incomplete. The cohort-wide trend, holding for low-risk and high-risk groups alike, points squarely at structural and environmental causes that operate on adolescents independent of their personal choices.
This is the epidemiological signature of an upstream cause. Whatever has been pulling adolescent sleep down over the past sixteen years has been pulling it down for everyone, regardless of what the individual teenager is doing in the evening. The authors recommend population-level rather than targeted interventions — delayed school start times are the canonical example.
The Dahl Foundation
The current sleep epidemiology rests on a body of developmental neuroscience built largely by Ronald Dahl and collaborators starting in the late 1990s. A foundational summary is Dahl and Lewin’s 2002 review in the Journal of Adolescent Health, which laid out the modern account of why adolescent sleep is what it is.
The headline biology: pubertal maturation shifts the circadian rhythm later. The shift is hormonal, not behavioral. Melatonin, the chemical that initiates sleep, gets released later in adolescents than in children or adults. Dim-light melatonin onset, measured in the lab, can shift by an hour or two during the pubertal transition. The shift is not a phase. It is the adolescent normal.
The practical consequence is that a developmentally typical teenager is not capable of falling asleep at 10 PM. The hormonal cascade that initiates sleep has not yet started. The teenager who is “still awake at 11” is not making a behavioral choice. The teenager’s brain has not yet released the chemical that initiates sleep.
Wake time matters too. The teenager whose alarm goes off at 6:30 AM is being asked to function on hormonal time-zones that match a roughly 8:30 AM wake. The chronic mismatch produces the cognitive and emotional dysregulation that gets attributed to laziness, attitude, screen addiction, or “this generation.” None of those attributions fit the biology. The biology fits a different attribution: this teenager has been sleep-deprived in a specific developmental-neuroscience sense for years.
What the Structural Cause Actually Is
If the cause is upstream, the question becomes: which upstream variables are doing the work? Four candidates worth naming.
School start times. The American Academy of Pediatrics formally recommended in 2014 that middle and high schools start at 8:30 AM or later, citing the developmental sleep literature. A 2022 meta-analysis of 28 studies covering 1.77 million students (Yip et al., Pediatrics) tightened the recommendation further: the 8:30-8:59 AM window is the start-time band with the most consistent association with longer sleep duration (r=0.109), less negative mood (r=0.060), and better aggregated developmental outcomes (r=0.061). Where the 7:30 start is in place, it is a structural sleep-deprivation policy applied to an entire population.
Activity stacking. The post-school window that was once discretionary has been compressed by sports, clubs, tutoring, college-prep programs, and part-time jobs. The biology says teenagers want to be active later; the schedule says their activity has to happen earlier.
Homework volume. Total homework load has increased over the same window where sleep has decreased. When the load is heavy, the 6 PM to midnight stretch that was once discretionary becomes the only time available for academic work. The cost is bedtime.
The phone-and-social-media accusation. The Bommersbach data show it cannot be the only cause, because adolescent sleep loss appears even in cohorts with low or no social media use. Phones contribute — they are a real source of arousal and attention-capture during the evening — but the structural causes are upstream. Removing the phone, without changing the schedule or the activity stack, recovers some sleep but does not solve the trend.
None of these is the kind of variable a parent can fix on a Tuesday night. They are the architecture of how American adolescence has been organized. The architecture was not designed with the developmental sleep literature in mind. It was designed for institutional convenience.
What Changes If We Take This Seriously
For schools: shift start times to align with adolescent biology. The Yip meta-analysis is honest about what changes and what does not. Districts that have delayed starts (the Seattle School District’s 2016 shift from 7:50 to 8:45 AM is the cleanest natural experiment in the literature) showed longer sleep and improved mood at the population level. Academic-outcome effects were null in the meta. Effect sizes are small, but at population scale for a low-cost structural change, they matter. The intervention disrupts adult schedules, which is why it has been rare. The pediatric sleep science says the disruption is worth it.
For parents: stop fighting the adolescent’s biology in the morning. Stop framing the kid’s struggle to wake as a character flaw. Reduce the activity-and-homework load where you have control over the load. Protect the evening window as the developmentally appropriate sleep window it is. The Saturday lie-in is not laziness. It is the body recovering the debt the week was running.
This is the part of the engineering failure parents have some agency over. The school start time and the policy-level conversation about adolescent biology are upstream variables that need to move at a different scale. What is in front of you on a Tuesday is the relationship with your own teenager — and the choice of whether to keep treating their biology as a behavior problem.
TRY THIS
The Saturday Diagnostic
For one week, do not wake your teenager any earlier than the school day requires. On Saturday, let them sleep until they wake on their own. Set no alarm. Do not knock on the door at 9 to ask about their plans. Do not vacuum outside their room at 10 to send a signal.
Notice how long they sleep. That is the sleep debt their system has been accumulating.
The number is the diagnostic. If your teenager wakes on their own at 8:30 AM on a Saturday after a normal week of 6:30 AM wake-ups, they were running close to enough. If they sleep until 11 or noon, the gap between what their biology needs and what their week has been giving them is being paid back, in real time, in your house, on a Saturday morning. That gap is debt, not laziness.
The diagnostic is not a one-time exercise. Run it for three or four Saturdays in a row to see the shape. A teenager whose Saturday wake time stays late week after week is telling you something about Monday-through-Friday that no amount of bedtime enforcement will fix. The bedtime is not the variable they have control over. The wake time, set by the school district, is.
CONSIDER THIS
What would change in your household if you accepted that your teenager’s biology will not align with 7 AM wake-ups until they are roughly 22 years old?
What would stop being a fight? What would start being a conversation? Which of the small daily rituals — the morning lecture, the eye-roll about the late wake, the Sunday-night anxiety about Monday — are downstream of treating the biology as a character problem? Some of those rituals are doing real damage to the relationship for no developmental benefit. Naming which ones is the move you can make this week. The school start time is the move someone else has to make.
Gabriele and I write more about parenting and developmental science at our Functional Child Development Substack.
Want the research behind this post? Read the Dive Deeper: The Engineering Failure — five studies that anchor what we wrote, what they found, and where the evidence has limits.
Every Functional Child Development post comes with a Dive Deeper companion. Subscribe for the full picture.