Jean Twenge

Who they are

Twenge works on generational data. Her method across four books is the same one: take large, repeated, nationally representative surveys that have asked the same questions for decades, and read the birth cohorts against each other rather than reading one cohort at a time. That design is what lets her say something about change over time that a cross-sectional study cannot, and it is also what makes her work argued about, because the same survey series will support several defensible analyses.

Her argument, put as she puts it. Something happened to American adolescents around 2012. Depressive symptoms rose, self-harm rose, and suicide rates among teenagers rose after two decades of decline, with the change concentrated among girls. The turn is sharp rather than gradual, it appears in more than one national survey, and it lines up with the year smartphone ownership passed fifty per cent in the United States. In the same period the things that had previously filled adolescent time, seeing friends in person, driving, working a summer job, dating, all fell. Adolescents who spend more hours on screens report worse mental health, and those who spend more hours on non-screen activities report better. She holds that a change this abrupt, this widely replicated across datasets and this well matched in timing is unlikely to be a coincidence, and that waiting for a perfect causal design before acting is itself a choice with costs attached to children.

The primary paper is Twenge, Joiner, Rogers and Martin (2018) in Clinical Psychological Science, drawing on two nationally representative surveys of US adolescents totalling 506,820 respondents together with national suicide statistics.

The objections are substantial and should be stated at their full strength rather than in summary. Amy Orben and Andrew Przybylski, running a specification curve across three datasets covering 355,358 people, found that digital technology use accounts for at most 0.4 per cent of the variation in adolescent wellbeing, marginally more than regularly eating potatoes. Przybylski's own first-authored work finds wellbeing rising with moderate use before it falls, so the relationship is a curve rather than a slope, and the harm at the heaviest use runs about a third as bad as missing breakfast. Christopher Ferguson reports a negligible relationship between screen time and depression. Sarah Rose Cavanagh's criticism is methodological rather than statistical, and is the one that costs most: that Twenge reviews the studies supporting her conclusion and not the ones that do not. Peter Gray adds a complication from another direction entirely, placing the decline in children's unstructured free play at around 1955, which puts a long trend seventy years ahead of the phone rather than downstream of it. Neither Cavanagh nor Gray has a page on this site yet.

There is a lineage worth naming. Nicholas Carr made a structurally similar case in 2010 from neuroplasticity, where Twenge made hers in 2017 from generational survey data. Both have had their headline evidence contested, and Nir Eyal argues against both. Reading the two together is instructive precisely because the arguments are independent of each other and arrive at the same place.

Where this site lands, September 2026. The correlation is real and the causal attribution is not established. Effect sizes in this literature are small and heavily moderated, which satisfies neither camp, and the honest description is a live dispute rather than a settled debunking. Twenge's core observation that something changed around 2012 stands. What would move us is natural-experiment evidence exploiting staggered smartphone or platform availability, which is the design the question actually needs and does not have. This is the most YMYL-adjacent disagreement in our register, it concerns children, and both camps are motivated, so it is the one where we hold the verdict most loosely.

Books

Four books, all trade rather than academic.

Her most-quoted single piece is not a book. "Have Smartphones Destroyed a Generation?" ran in the Atlantic in September 2017, and that headline has done a great deal of the arguing on her behalf ever since. The article underneath it is more qualified than the title, which is worth knowing before quoting either.

Key papers

Twenge, J. M., Joiner, T. E., Rogers, M. L., & Martin, G. N. (2018). "Increases in Depressive Symptoms, Suicide-Related Outcomes, and Suicide Rates Among U.S. Adolescents After 2010 and Links to Increased New Media Screen Time." Clinical Psychological Science, 6(1), 3-17. First author, and the primary paper behind the argument. Two nationally representative surveys of United States adolescents, more than half a million respondents between them, read alongside national suicide statistics for ages thirteen to eighteen. Depressive symptoms, suicide-related outcomes and suicide rates all turn upward after 2010, with girls worst affected.

Twenge, J. M. (2017). "Have Smartphones Destroyed a Generation?" The Atlantic, September 2017. Not a paper, and the single most quoted thing she has written. The article is considerably more qualified than its headline, which is worth knowing before either is cited.

The papers on the other side, which belong in any honest reading of this dispute. Orben, A., & Przybylski, A. K. (2019), Nature Human Behaviour, 3, 173-182, a specification curve across three datasets covering more than 350,000 people, finding that digital technology use explains at most a fraction of one per cent of the variation in adolescent wellbeing. Przybylski, A. K., & Weinstein, N. (2017), Psychological Science, 28(2), 204-215, finding wellbeing rising with moderate use before it falls, so a curve rather than a slope. Ferguson, C. J. (2017), Psychiatric Quarterly, 88(4), reporting a negligible relationship between screen time and depression.

The trend is not in dispute between these papers. The attribution and the magnitude are, and the design that would settle it, a natural experiment exploiting staggered smartphone or platform availability, has not been run.

Key ideas

The 2012 inflection: multiple independent national surveys show adolescent mental health turning sharply downward in the same short window, with girls worst affected.

The displacement account: hours moved from in-person socialising, driving, paid work and dating to screens, and the two sets of activities point in opposite directions in the data.

Cohort rather than age: reading birth cohorts against each other separates generational change from the ordinary business of growing up.

Her position on evidentiary thresholds, which is a value judgment rather than a finding: where the subject is children, waiting for causal certainty is not a neutral act.

The contested part is attribution and magnitude, not the trend. Her critics agree something changed.

Concepts on this site

Techno-Panic, the pattern her argument is the current entry in, and which she would say the pattern's existence does not refute

Digital Loneliness, where the two readings of the evidence are set out at length

Behavioral Addiction

---