There is one chart that has done more to shape how a generation of parents thinks about adolescence than any study, and it is worth understanding properly rather than through the argument about it.
Jean Twenge is a professor of psychology at San Diego State University, and she has spent her career on a particular kind of question: not how individuals differ from each other, but how one birth cohort differs from the last. That is a harder question than it sounds, because answering it requires the same measures administered to new teenagers year after year for decades. The United States happens to have those datasets, among them Monitoring the Future and the CDC's surveys of adolescent health, and Twenge has read them more closely than almost anyone.
What she found, and published in iGen in 2017 and in The Atlantic that year under the title "Have Smartphones Destroyed a Generation?", is that several indicators of adolescent wellbeing turned at roughly the same moment. Loneliness. Depressive symptoms. Time spent with friends in person. And the turn lands in the early 2010s, around 2012, which is also when smartphone ownership among American teenagers crossed from a minority to a majority.
Her argument is that the coincidence is not a coincidence.
Take the strongest version of her case rather than the caricature, because the strong version deserves an answer. It is not merely that two lines moved together. It is that the turn appears across multiple independent datasets rather than one. It appears in behaviors as well as in self-reported feelings, which matters because behaviors are harder to explain away as a change in how willing young people are to report distress. It appears in several countries. And the timing is specific rather than vague: not a slow drift across the decade, but a bend in the curve at a point when the technology of adolescent social life changed in a discrete way.
That is the case. It is a serious one, and anyone who dismisses it as technophobia has not read it.
Here is where I have to be careful, and so should anyone building on this.
Her method is correlational, and her reading of the correlation is exactly what other serious researchers dispute. Amy Orben at Cambridge and Andrew Przybylski at Oxford, among others, have argued that when you examine the same kinds of dataset with different analytic choices, the association between digital media use and adolescent wellbeing is extremely small, comparable in size to associations nobody worries about at all, and that the conclusion is sensitive to which variables you include. That is not a quibble about statistics. It is a disagreement about whether the effect is large enough to be the explanation for a generational turn.
Both of those researchers appear on our Evidence page, deliberately, alongside Twenge. A reader who wants the other side will find it there rather than having to look for it.
So what survives the disagreement, and can be built on?
Something did change in the early 2010s in the social world of adolescence. That much is not in dispute by anyone, in either camp. What is in dispute is how much of the wellbeing change it caused.
And this is the third book in The Unmistakably You Series, You, Not a Copy, which is about the mold: the pressure to produce an acceptable version of a self before anybody has met the real one. For that argument, the contested part is not actually load-bearing.
Because the change Twenge's timing points at is not only a change in hours. It is a change in whether an adolescent's social life has an audience, a record and a scoreboard. A fourteen year old in 2008 performed a version of herself for the people in the corridor and the performance ended when she went home. A fourteen year old in 2016 performed it for an unbounded audience, permanently, with a number attached.
You can be entirely agnostic about the depression curve and still notice that the conditions for self-presentation changed completely, in a few years, for everyone that age at once. Whether it moved the clinical numbers is a genuine open question. Whether it moved the incentives to edit yourself is not really arguable.
There is one more thing in her favor that the argument about the 2012 chart tends to obscure. Twenge did not begin as a researcher of screens. Her method is cohort comparison itself, applied across the whole twentieth century, and her 2023 book Generations takes the same datasets and asks what distinguishes every living generation from the one before it, including on questions that have nothing to do with technology. That matters because it means the 2012 claim is not an ad hoc explanation reached for to fit one worrying curve. It comes out of a long-standing method with a track record on questions where nobody had a stake in the answer.
One chart started the argument. The argument is still live, and the part everyone agrees on was always the more interesting half.
Want to explore these ideas further? Learn more about my work on leadership, identity and conviction at www.theglennllopis.com.
This essay was first published on Conviction Chronicles

