Verified by 
“Have smartphones destroyed a generation?” In 2017, psychologist Jean Twenge penned a magazine article with that provocative question in the headline. Drawing on her own research, Twenge’s article (an excerpt from her book iGen) painted a grim picture of a generation on “the brink of the worst mental health crisis in decades.” Twenge, who studies the generation of adolescents and young adults born between 1995 and 2012—known as Generation Z or iGen—posited that increasing use of digital technology, and particularly social media, might be to blame.
Twenge and her colleagues’ findings—that the more teens used digital technology, the more likely they were to be depressed, suicidal, and sleep-deprived—were covered widely, and campaigns to restrict children’s access to smartphones or to encourage teens to limit their own social media use have gained some ground. Yet some in the field criticized the researchers for “cherry-picking” only the data that supported their hypotheses and for employing flexible analytic methods that made any statistically significant effect appear socially devastating. Now, two new studies—one using some of the same data that Twenge has—are calling her theory into question.
The first paper, published last month in Nature Human Behavior, examined data from three large-scale surveys: Monitoring the Future and the Youth Risk and Behavior Survey, both from the U.S., and the Millenium Cohort Study, from the U.K. Together, the datasets covered more than 350,000 teenagers between the ages of 12 and 18; all were surveyed between the years 2007 and 2016. (Monitoring the Future, which is funded by the National Institute on Drug Abuse, was the basis for much of Twenge’s research for iGen.) The three surveys ask participating teens various questions about digital technology use, mental health, dietary habits, and other aspects of their wellbeing.
Since each survey asked many different questions of its participants, the authors note, there were more than 60,000 ways in total that the researchers could have analyzed the relationship between digital technology use and teen mental health. Rather than looking at one associative pathway between wellbeing and tech use, as past research has done, the authors of the new paper used a method called Specification Curve Analysis (SCA) to examine every possible analytic pathway and get an overall snapshot of how technology and teen mental health interact with one another.
The technique is designed to minimize the effects of researcher bias, says Amy Orben, a Ph.D. candidate at the University of Oxford who co-authored the paper with Oxford psychologist Andrew Przybylski. “Small biases that researchers have—either unconscious or conscious—can skew the results they find when analyzing a dataset,” she says. “[SCA] tries to look at the diversity of data analysis. Instead of running just one analysis, we went down every single possible path. It’s as if we simulated that there were 20,000 research teams, all with their own biases and histories, and looked at the range of possible results they all could’ve found.”
They found that, when all possible outcomes were taken into account, digital technology use—which included social media use, TV viewing, and using the internet to keep up with news—was negatively associated with teen wellbeing. But the association was extremely small: Digital tech use explained at most 0.4 percent of variation in wellbeing across the large samples. By comparison, eating potatoes regularly had a similar negative association with teen mental health; wearing glasses was more negatively associated with poor wellbeing than tech use. “What this shows is that previous claims of intense negative effects of digital technologies on teen wellbeing were built on very unstable foundations,” Orben says.
The data examined for the current research, including those used by Twenge, say nothing about causality, she notes. “[This study] is not saying that potatoes cause your child to feel bad—just as it’s not saying that social media causes your child to feel bad.” There may be other, unexamined factors that are contributing to poor mental health among teens, she says, and certain demographics or individuals may be more vulnerable than others. Ultimately, further research may determine that digital technology use does have a strong negative effect on adolescent wellbeing, she adds. “But we’re only at the very beginning of figuring out what these associations actually are.”
The question of causality, in particular, “has always plagued this field of research,” says Christopher Ferguson, a psychologist at Stetson University who was not involved in the study. Correlations are often incorrectly communicated to the public as instances of cause and effect, he adds, particularly on “emotive” topics like teen mental health or the rise of smartphone use. “As we get older, we become more and more suspicious of new technology,” he says. “A lot of people are already primed to want to believe negative things about [social media], and therefore headline-grabbing claims of impending doom tend to get more attention.”
He notes that it’s necessary to ask the question of directionality: “Is it that kids use screens and then have more mental health problems—or is it that kids who have more mental health problems tend to use more screens?”
Though Twenge’s work suggests the former pathway, another recent study provides evidence for the latter. Published last month in Clinical Psychological Science, the study used a longitudinal approach, examining how social media use and depressive symptoms changed over time in a sample of adolescents (surveyed each year for two years) and another sample of young adults (surveyed over six years).
Social media use did not predict later depressive symptoms in either sample. The reverse pathway—depressive symptoms at one time predicting increased social media use at a later time—was only observed in adolescent girls.
The study, subtitled “An Empirical Reply to Twenge et al,” was conceived after the authors read a paper co-authored by Twenge in 2018, says lead author Taylor Heffer, a Ph.D. student at Brock University in Canada. “We realized we were able to address some of the concerns [other researchers] had raised, and we had two longitudinal samples to test Twenge’s hypothesis that greater social media use might be associated with more depressive symptoms over time.”
Heffer emphasizes that even though they were able to look more clearly at directionality than Orben and her collaborators, she and her co-authors were still unable to establish causality. “Even with longitudinal designs, it’s possible that there could be other overlooked factors that might be responsible for the associations,” she says. Still, she adds, “Our study highlights that fear of social media use may be premature.”
Twenge reports that she is working on a response to the Nature Human Behavior paper. She says that while the variance in mental wellbeing observed by the researchers was small, it could have real-world implications that are being downplayed by Orben and her co-authors. “Using the same data, those who spend 5+ hours a day on devices—compared to less than 1 hour a day—are more than twice as likely to have attempted suicide,” she says. “I can't see how that can possibly be seen as small or as having no practical importance.”
Ferguson notes that because both studies were based on self-reported data—both for technology use and for mental wellbeing—it’s possible that the findings aren’t perfect representations of the associations that exist.
Despite that limitation, which exists for almost all the research in the field, he adds, the two studies put the larger conversation into perspective. “What we’re seeing from these studies is that, even if the results may be statistically significant in some large samples, that they’re no bigger than other effects that we don’t take seriously,” he says. “We’re not running around saying potatoes are causing teens to commit suicide.”
Attempts to verify research on the topic of digital technology use and mental health are important in the context of psychology’s ongoing replication challenges, Ferguson adds. “But I think the next challenge is going to be admitting that some of our statistically significant effects are nonetheless not worthy of informing the public about. We can publish them, but we ought to interpret them very cautiously and conservatively. Otherwise, we’re misleading people.”
