In September 2017, Jean Twenge published a cover article in The Atlantic under the headline "Have Smartphones Destroyed a Generation?" It opened with a vivid demographic portrait of post-millennial teenagers who had grown up with iPhones — a generation she called iGen — and argued that the devices were causing an epidemic of loneliness, depression, and anxiety. The data behind the claim were real: across multiple large national surveys, rates of teenage depression, anxiety, self-harm, and suicide had risen sharply since around 2012, and girls had been affected more than boys, and the inflection point matched the moment when smartphone ownership among adolescents crossed 50%.
The article was read millions of times. Two years later, Amy Orben and Andrew Przybylski published a methodological response in Nature Human Behaviour that became famous for a single comparison. Analyzing two large datasets — the UK Millennium Cohort Study and the US Monitoring the Future survey — using specification curve analysis that tested thousands of analytical choices simultaneously, they found that the association between social media use and adolescent well-being was real but very small. In fact, they calculated, the association between wearing glasses and adolescent well-being was similarly sized. So was the association between eating potatoes and adolescent well-being. The finding did not mean social media had no effect. It meant the claimed effect was too small to explain a large population-level trend, and that the certainty with which the smartphone narrative had been presented was not supported by the data.
That tension — between a real and serious trend in adolescent mental health and contested, complicated, and politically charged explanations for it — is the central question this article examines. Understanding what has actually happened to mental health rates, and what has caused it, requires separating genuine signal from measurement artifact, evaluating competing causal claims with appropriate rigor, and grappling with the possibility that the answer is not one thing.
"Adolescent girls are not fragile. But they are living in a world that has become measurably more stressful in ways that are not primarily of their own making." — Candice Odgers, Nature (2018)
Key Definitions
Prevalence: the proportion of a population with a given condition at a specific point in time; distinct from incidence (the rate of new cases).
Diagnostic expansion: the broadening of criteria for a psychiatric diagnosis over successive editions of diagnostic manuals, which increases measured prevalence without necessarily increasing true prevalence.
Measurement artifact: an apparent change in prevalence that reflects changes in how a condition is measured, diagnosed, or reported rather than changes in its actual occurrence.
Youth Risk Behavior Survey (YRBS): the US Centers for Disease Control and Prevention's biennial survey of high school students tracking health-related behaviors including mental health symptoms; one of the primary data sources for adolescent mental health trends.
Specification curve analysis: a methodological technique that tests all (or a large sample of) reasonable analytical choices simultaneously to assess the robustness of a finding across different model specifications.
Social comparison: the process of evaluating one's own qualities, opinions, and circumstances relative to those of others; a mechanism through which social media use may affect mental health.
Passive social media use: scrolling through content without posting or interacting; associated with more negative mental health outcomes than active, interactive use in research.
Monotropism: a theory of autism spectrum cognition; not directly relevant here, but cited in some research examining differential mental health effects of social media across diagnostic groups.
Internalizing symptoms: psychological distress directed inward, including depression, anxiety, and withdrawal; more common in girls; contrasted with externalizing symptoms (aggression, conduct problems) more common in boys.
What the Epidemiology Actually Shows
The empirical question of whether adolescent mental health has genuinely deteriorated requires distinguishing several different phenomena that are often conflated.
Changes in Subjective Distress
The CDC's Youth Risk Behavior Survey has tracked self-reported feelings of persistent sadness, hopelessness, and suicidal ideation since 1991 using consistent methodology. This consistency is its primary strength for trend analysis: if the questions and sampling methods have not changed substantially, genuine changes in responses probably reflect genuine changes in experience rather than measurement changes.
The YRBS shows a clear pattern: from 1991 to approximately 2011-2012, rates of persistent sadness and hopelessness in high school students were roughly stable at around 28-30%. From 2011-2012 onward, these rates increased substantially. The 2021 YRBS found that 44% of high school students reported persistent feelings of sadness or hopelessness in the past year — a 13-percentage-point increase from a decade earlier. Rates were substantially higher for girls (57%) than boys (29%). Rates among LGBTQ+ students were dramatically elevated (69%).
Changes in Severe Outcomes
Suicide and self-harm are harder to attribute to diagnostic expansion or changed reporting — they are behavioral outcomes recorded by hospitals and vital statistics. Suicide rates among adolescent girls aged 10-14 doubled between 2007 and 2015. Emergency department visits for self-harm among girls aged 10-14 increased substantially in the early 2010s. These are the hardest outcomes to attribute to artifact, and they show the same directional trend as self-reported symptoms.
The Role of Diagnostic Expansion
The DSM-5, published in 2013, broadened criteria for several conditions and introduced new diagnostic categories. Primary care physicians now screen more routinely for depression and anxiety. Reduced stigma around mental health has made it easier for people to acknowledge and seek help for symptoms. These changes do increase measured prevalence without necessarily increasing suffering, and they complicate trend analysis for conditions where diagnostic criteria have changed.
For adolescent depression specifically, however, the most commonly used survey measures in national trend data — which ask directly about symptoms (sad mood, loss of interest, sleep problems) rather than diagnostic categories — are less susceptible to this artifact. The trends appear in both diagnostic rates and symptom scales, suggesting genuine increases in distress rather than purely artifact.
The Smartphone Hypothesis: Evidence and Critique
Jean Twenge's case for smartphones as the primary cause of rising adolescent mental health problems rests on several observations: the timing alignment between smartphone adoption (reaching 50% of US adolescents around 2012) and the mental health inflection; the greater effect on girls, who use social media more intensively; the displacement of in-person social activity by screen time; and the association between heavy social media use and worse mental health in survey data.
The mechanism proposed involves several pathways: social comparison through image-centric platforms like Instagram reducing self-esteem and body satisfaction; cyberbullying through always-available digital channels; sleep disruption through nighttime phone use; and displacement of in-person social activity that provides more reliable sources of belonging and self-worth.
The empirical support is mixed. Experimental studies provide the strongest causal evidence. Hunt and colleagues' 2018 study in the Journal of Social and Clinical Psychology found that limiting social media use to 30 minutes per day for three weeks reduced loneliness and depression in college students compared to a control group. Studies in which participants deactivate Facebook or Instagram for several weeks show mood improvements in the deactivation group. These suggest a real causal effect.
The Orben-Przybylski Challenge
Orben and Przybylski's 2019 specification curve analysis found that the effect size of social media use on adolescent well-being in large observational datasets was approximately 0.05 standard deviations — a statistically detectable but substantively tiny effect. Their potato comparison illustrated the point dramatically: correlations of similar magnitude exist between well-being and countless other variables. This does not mean the association is zero, but it means it is too small to drive a large population-level trend.
Candice Odgers' Structural Critique
Candice Odgers, a developmental psychologist, has raised a more fundamental challenge: if smartphones were causing the mental health crisis, the effects should be largest among the groups with greatest smartphone exposure. But low-income and minority teenagers, who have higher baseline mental health burden, tend to have less and later access to expensive smartphones and faster internet connections. The pattern does not fit the hypothesis that more smartphone exposure causes more harm.
Odgers also notes that the largest increases in adolescent mental health problems have occurred in regions and populations where smartphone penetration is comparatively low, and that cross-national data do not show a consistent relationship between smartphone adoption rates and mental health trends.
The COVID-19 Effect
The COVID-19 pandemic produced the largest single documented increase in adolescent mental health problems in recent data. School closures removed the primary institutional source of social connection for adolescents. Social isolation during lockdowns was acute. For many young people, especially those in difficult home environments, lockdowns represented genuine psychological harm.
The 2021 YRBS data, collected during the pandemic period, show the largest single-survey increases in reported sadness, hopelessness, and suicidal ideation in the survey's history. The American Academy of Pediatrics declared a national emergency in child and adolescent mental health in October 2021. Meta-analyses of studies conducted during COVID-19 lockdowns find elevated rates of depression and anxiety in children and adolescents compared to pre-pandemic norms.
The pandemic is important not only as a cause of distress but as a proof of concept: clearly identifiable stressors (isolation, school closure, economic hardship, parental job loss, disease fear) do cause measurable increases in adolescent mental health problems. This is consistent with the view that other, less acute stressors — economic precarity, climate anxiety, reduced social infrastructure — may contribute to longer-term trends through similar mechanisms.
Non-Screen Explanations
Several structural factors have been proposed as contributors to the adolescent mental health trend that receive less media attention than smartphones but have comparable or stronger empirical support.
Economic Precarity and Student Debt
Young people in wealthy countries today face measurably worse economic prospects than their parents did at the same age in important respects: higher housing costs relative to income, greater student debt loads, more precarious employment with fewer benefits, and delayed milestones (homeownership, family formation) associated with stable adult identity. Financial insecurity is one of the most robust predictors of depression and anxiety in longitudinal research. The stagnation of real wages for workers without college degrees, concentrated precisely in the age range of adolescents' parents, creates household-level stress that affects children.
Climate Anxiety
A 2021 Lancet Planetary Health study surveyed 10,000 young people aged 16-25 across ten countries and found that 59% were very or extremely worried about climate change, 48% said it negatively affected their daily functioning, and over 50% reported feelings of sadness, anxiety, anger, powerlessness, helplessness, and guilt related to climate change. This is a novel stressor with no historical precedent for the generation experiencing it: the prospect of severe and worsening climate disruption over the course of their lifetimes.
Reduced Social Infrastructure
Robert Putnam's research on social capital — the networks of civic engagement, community belonging, and interpersonal trust that support individual well-being — documented its decline across American society since the 1970s. Churches, civic organizations, bowling leagues, and neighborhood associations that once provided belonging and meaning have declined in membership and participation. For young people, the erosion of community institutions that once structured adolescence outside of school represents a loss of belonging infrastructure that is not replaced by digital connection.
The Haidt-Lukianoff Thesis
Jonathan Haidt and Greg Lukianoff's 2018 book The Coddling of the American Mind argues that overprotective parenting, safe space culture, and institutional risk-aversion have reduced young people's tolerance for adversity and their capacity to cope with setbacks. This thesis has attracted both significant popular resonance and sharp academic criticism. Critics argue it blames young people for structural conditions, lacks strong empirical grounding, and overestimates the generality of campus social trends. The thesis may capture real dynamics in some elite educational contexts while poorly describing the experiences of most young people.
Gender Differences
The gender disparity in the adolescent mental health trend is among its most robust features and most contested explanatory challenges. Depression and anxiety rates have risen for both sexes but substantially more for girls. In the 2021 YRBS, 57% of girls reported persistent sadness or hopelessness compared to 29% of boys. This gap has widened over the trend period.
Several mechanisms are proposed. Image-centric social media platforms may affect girls more because appearance-based social comparison is a stronger driver of self-esteem for girls than boys. Cyberbullying tends to be more relational and reputational in form among girls, which may cause more harm. Girls' stronger social orientation may mean that disrupted peer relationships — both through social media conflict and through COVID isolation — cause greater distress.
But the gender gap predates the smartphone era. Girls have shown higher rates of depression and anxiety than boys from puberty onward across decades of epidemiological data, observed cross-culturally. Early puberty in girls is an established risk factor for depression. Academic pressure affects girls particularly acutely: girls significantly outperform boys in academic achievement while also reporting higher stress. These factors predate smartphones and suggest that the technological explanation captures only part of the gender differential.
What Interventions Work
Evidence-based interventions for adolescent mental health operate at multiple levels with variable quality of evidence.
Clinical Treatment
Cognitive-behavioral therapy (CBT) has the strongest evidence base for adolescent depression and anxiety: meta-analyses find effect sizes of 0.5 to 1.0 standard deviations compared to control conditions, maintained at follow-up. Behavioral activation, mindfulness-based therapies, and interpersonal therapy also have supporting evidence. The challenge is access: the great majority of adolescents with diagnosable depression and anxiety receive no treatment. Therapist shortages, cost, stigma, and geographic barriers limit access severely.
School-Based Programs
Universal and targeted school-based mental health programs that teach emotional regulation, cognitive restructuring, and problem-solving skills have shown modest positive effects in trials. Implementation quality varies substantially, and effects are often smaller in real-world settings than in efficacy trials. They can reach young people before problems become severe, which is their primary advantage over clinical treatment.
Platform-Level Interventions
Experimental platform interventions — removing like counts, changing algorithmic content curation, age verification, use-time limits — have theoretical support but limited rigorous evaluation at population scale. Australia's 2024 law restricting social media for children under 16 represents a natural experiment whose effects will be studied. Limiting passive consumption and social comparison features appears more promising than blanket restrictions, based on the mechanism evidence.
Structural Interventions
Reducing student debt, improving housing affordability, providing economic security for young people, and addressing the material conditions associated with economic precarity would likely produce mental health benefits. Climate policy that creates a sense of agency and reduces helplessness may reduce climate anxiety specifically. These interventions require political will and operate outside the mental health system, but the evidence base for their relevance is strong.
Connections
For the specific evidence on depression's causes and treatments, see what causes depression. For anxiety disorders specifically, see what causes anxiety. For how social media platforms affect brain and behavior beyond mental health, see how social media rewires the brain.
References
Twenge, J. M. (2017, September). Have smartphones destroyed a generation? The Atlantic. Retrieved from theatlantic.com
Orben, A., & Przybylski, A. K. (2019). The association between adolescent well-being and digital technology use. Nature Human Behaviour, 3(2), 173-182. https://doi.org/10.1038/s41562-018-0506-1
Centers for Disease Control and Prevention. (2022). Youth Risk Behavior Survey: Data Summary and Trends Report, 2011-2021. CDC.
Odgers, C. L. (2018). Smartphones are bad for some teens, not all. Nature, 554(7693), 432-434. https://doi.org/10.1038/d41586-018-02109-8
Hunt, M. G., Marx, R., Lipson, C., & Young, J. (2018). No more FOMO: Limiting social media decreases loneliness and depression. Journal of Social and Clinical Psychology, 37(10), 751-768. https://doi.org/10.1521/jscp.2018.37.10.751
Hickman, C., Marks, E., Pihkala, P., et al. (2021). Climate anxiety in children and young people and their beliefs about government responses to climate change. The Lancet Planetary Health, 5(12), e863-e873. https://doi.org/10.1016/S2542-5196(21)00278-3
Haidt, J., & Lukianoff, G. (2018). The Coddling of the American Mind: How Good Intentions and Bad Ideas Are Setting Up a Generation for Failure. Penguin Press.
Rao, W.-W., Zong, Q.-Q., Zhang, J.-W., et al. (2021). Prevalence of depression in children and adolescents during COVID-19: A meta-analysis. EClinicalMedicine, 38, 101041. https://doi.org/10.1016/j.eclinm.2021.101041
Frequently Asked Questions
Are mental health rates actually increasing or are we just diagnosing more?
Both phenomena are happening, and separating them is genuinely difficult. On one hand, diagnostic expansion is real: the DSM-5 (2013) broadened criteria for several conditions, primary care physicians now screen more routinely for depression and anxiety, and reduced stigma has made it easier for people to acknowledge and seek help for mental health symptoms. All of these factors increase measured prevalence without necessarily meaning more suffering.On the other hand, multiple indicators that are harder to attribute solely to diagnostic change suggest real increases in distress among adolescents, particularly girls. The CDC's Youth Risk Behavior Survey (YRBS), which has tracked self-reported feelings of sadness, hopelessness, and suicidal ideation since the early 1990s using consistent methodology, shows clear inflections beginning around 2011-2012. In 2021, 44% of high school students reported persistent feelings of sadness or hopelessness — a 13-percentage-point increase from 2011. Emergency department visits for self-harm among adolescent girls increased substantially in the early 2010s. Suicide rates for girls aged 10-14 doubled between 2007 and 2015 — an increase in the hardest possible outcome to attribute to diagnostic artifact.Cross-national data also support real change. Similar inflections appear in the UK, Canada, Australia, and several European countries at roughly the same time, suggesting a common cause rather than purely American diagnostic trends. The balance of evidence suggests both things are true: genuine increases in adolescent distress since roughly 2012, amplified in measured rates by increased recognition and treatment-seeking. The honest uncertainty lies in exactly how large the genuine increase is, what caused it, and whether trends differ by severity — the most serious outcomes (self-harm, suicide) may have increased more than mild or moderate symptoms.
What role do smartphones play in adolescent mental health?
The smartphone hypothesis, most prominently advanced by Jean Twenge in her 2017 Atlantic article and her book iGen, argues that the widespread adoption of smartphones — particularly from around 2012, when US smartphone ownership crossed 50% — caused the rise in adolescent depression and anxiety by displacing in-person social activity, exposing teenagers to social comparison on social media, disrupting sleep through nighttime use, and creating new vectors for cyberbullying.The hypothesis has intuitive appeal because the timing fits: the mental health inflection tracks smartphone adoption closely, and the effect is larger for girls, who tend to use social media more and show more social-comparison vulnerability. Experimental studies show that heavy Instagram use among teenage girls increases body dissatisfaction and depressive symptoms. Meta-analyses of cross-sectional data consistently find associations between heavy social media use and worse mental health outcomes.But the causal picture is sharply contested. Amy Orben and Andrew Przybylski's 2019 paper in Nature Human Behaviour analyzed large UK and US datasets using specification curve analysis — testing thousands of analytical choices simultaneously to assess robustness — and found that the association between social media use and adolescent well-being is real but tiny. Their most-cited finding: the effect size is similar to the association between wearing glasses and well-being, or between eating potatoes and well-being. Correlations exist but are too small to explain large population-level trends.Candice Odgers has further argued that the smartphone narrative may be misleading because low-income and minority teenagers, who have the highest mental health burden, tend to have less and later smartphone access — the opposite pattern from what a causal story would predict. The current scientific consensus, if one exists, is that heavy social media use probably contributes to distress for vulnerable individuals (particularly girls prone to social comparison), but is unlikely to be the primary driver of population-level trends.
Why are girls more affected than boys?
The gender disparity in adolescent mental health trends is one of the most robust findings in this literature and one of the most actively debated. Depression and anxiety rates have risen for both sexes, but the increase has been substantially larger and steeper for girls and young women. In the US, rates of major depressive episodes among adolescent females roughly doubled between 2011 and 2021 according to the National Survey on Drug Use and Health, while rates for males increased more modestly.Several mechanisms are proposed. Social comparison through image-centric social media platforms (Instagram, TikTok) may affect girls more because appearance-based social comparison is a stronger driver of self-esteem in girls, and because algorithm-driven feeds preferentially surface idealized body and lifestyle content. Cyberbullying, which tends to be more relational and reputational in form among girls than among boys, may also cause more harm. Jonathan Haidt and Jean Twenge have emphasized these social-media-specific mechanisms.But the gender gap predates the smartphone era. Girls show higher rates of depression and anxiety than boys from puberty onward — a well-established pattern going back decades and observed cross-culturally. Puberty itself, particularly early puberty in girls, is a risk factor for depression. This suggests underlying biological and social vulnerabilities that long predate smartphones, which the smartphone era may have amplified rather than created.Other factors are not gender-neutral: girls report higher levels of climate anxiety, higher stress from academic pressure (grade trends show girls significantly outperforming boys in academic achievement but reporting higher stress), and greater exposure to sexual harassment online and offline. The gender disparity is probably multiply determined — no single cause explains it, and interventions addressing only social media use while ignoring other structural factors are likely to have limited effect.
What other factors explain the trends?
Beyond smartphones, researchers have proposed a range of structural and social explanations for rising mental health distress, several of which have better empirical support than the social media hypothesis.Economic precarity is consistently associated with mental health outcomes, and young adults in wealthy countries today face worse economic prospects than their parents did at the same age in several measurable ways: higher housing costs relative to income, greater student debt, less job security, and delayed family formation. These stressors are real and measurable. Longitudinal studies consistently show that financial insecurity predicts depression and anxiety.Climate anxiety represents a newer but growing stressor, particularly among young people who report high levels of concern about climate change and feelings of hopelessness about the future. Survey data from multiple countries find that a substantial minority of young people report climate change affecting their daily mental health, with a 2021 Lancet study finding that 59% of young people aged 16-25 across ten countries reported being very or extremely worried about climate change.Political polarization and chronic background stress from a media environment of continuous crisis coverage — mass shootings, pandemics, democratic backsliding — may contribute to generalized anxiety in ways that are difficult to separate from other causes. Jonathan Haidt and Greg Lukianoff's 'safetyism' thesis in The Coddling of the American Mind argues that overprotective parenting and risk-averse institutional cultures have reduced young people's tolerance for adversity, but this hypothesis has attracted criticism for its methodological basis and for blaming individuals rather than structural conditions.Decline in religious participation and community belonging — loneliness is a robust predictor of depression — may also contribute. Robert Putnam's research on declining social capital, extended to younger generations, suggests that the erosion of institutions that provide belonging and meaning has costs for mental health that are easy to overlook in analyses focused on technology.
Does social media cause depression?
The causal question is harder to answer than the correlational one. Cross-sectional studies — asking people about their social media use and their mental health at the same time — consistently find associations, but cannot determine direction: do unhappy people use social media more, or does social media use make people unhappy? Both are plausible and likely both operate.Experimental studies offer cleaner evidence. Deactivation studies — randomly assigning participants to deactivate Facebook or Instagram for several weeks — show mood improvements in the deactivation group, suggesting some causal effect. Hunt et al. (2018) found that limiting social media to 30 minutes per day for three weeks reduced loneliness and depression in college students. Jeffrey Hancock and colleagues' 2019 meta-analysis found a small but significant causal effect of social media use on well-being in experimental studies, smaller than in correlational studies.Mechanism matters for understanding magnitude. Passive consumption — scrolling through others' curated highlight reels without posting or engaging — appears more harmful than active communication and interaction. Upward social comparison, exposure to idealized body images, and fear of missing out (FOMO) are proposed mechanisms with some experimental support. Cyberbullying is a well-documented harm with clear causal evidence.The overall picture is that social media use does causally contribute to depression and anxiety for some users, particularly adolescent girls engaging in passive social comparison. But effect sizes in experimental and longitudinal studies are consistently smaller than in cross-sectional correlations, and smaller than the public discourse suggests. Social media is probably one contributing factor among many rather than the primary driver of population-level trends.
What interventions actually help?
Evidence-based interventions for adolescent mental health operate at multiple levels, from individual clinical treatments to policy and structural changes, with varying quality of evidence.At the individual and clinical level, cognitive-behavioral therapy (CBT) has the strongest evidence base for both depression and anxiety in adolescents: multiple meta-analyses find effect sizes of 0.5 to 1.0 standard deviations compared with control conditions, with effects maintained at follow-up. Behavioral activation, mindfulness-based therapies, and interpersonal therapy also have supporting evidence. The challenge is access: most young people with diagnosable depression and anxiety receive no treatment, and treatment capacity is severely constrained in most health systems.School-based mental health programs can reach young people before problems become severe. Programs that teach emotional regulation, CBT skills, and problem-solving have shown positive effects in trials, though effects are often modest and implementation quality varies enormously. Universal screening in primary care settings increases identification and treatment access.At the platform and policy level, the evidence for specific interventions is weaker but growing. Removing like counts (tested by Instagram), reducing algorithmic amplification of comparison content, and implementing age verification for high-intensity platforms have theoretical support but limited rigorous evaluation. Australia's 2024 legislation banning social media for children under 16 represents a population-level experiment whose effects will be studied.Addressing structural factors — reducing student debt, improving housing affordability, providing economic security for young people — would likely produce mental health benefits but requires political will beyond the scope of mental health policy. Climate policy that creates a sense of agency and reduces helplessness among young people has been proposed as a mental health intervention, though evidence at the population level is speculative. The lesson from the evidence is that no single intervention is sufficient; the most effective approaches combine accessible clinical treatment, preventive programs, platform accountability, and structural economic support.