I'm not a social scientist. I'm a middle school teacher. Based on my own anecdotal experience, I would reckon that social media is having a profoundly deleterious cultural effect that is difficult to measure and is collective rather than simply individual (and pscyhological symptomatic). There has never before been a human culture in which the young were allowed to exist in their own ideational spaces in this way, deluged with information and ideas and images and emotions crafted by manipulators and groomers and antisocial actors and profit seekers.
One of the kindergarteners in my school had incest domination porn on his iPad. A 3rd grader had BDSM material. If you're interested in the type of school I teach at you can subscribe to my Substack (or read my most recent post) but that should barely matter. Kids shouldn't be marinating in this content. As a formerly antisocial, impulsive, addiction-prone boy myself I know that this new technological landscape would have ruined my life. I have very little doubt. Where are the adults? As teachers we can't intervene. Parents won't. The bureaucracy (counselors, psychologists, healthcare providers, social workers, etc.) doesn't seem to be too concerned by it all. Perhaps it's just anticipating the wave of dysfunctional clients to come (which means resources and career advance and budget increases and public credibility). Just give it 5-10 years...
I don't think that social media use by kids is going to last. There's more and more restrictions being put up. All you have to do is look at countries like Australia and the UK. If anything, I believe that it's going to get to the point where AI itself starts putting up restrictions, as AI is getting more powerful, even as it's being fed loads of info that social media use by the youth is harmful.
Having said the above, I do share the views of Mike Males, that there's other factors, including negative parenting, that's lowering the mental health of young people. But in a good chunk of these cases, it's the system that's causing negative parenting especially in regards to Ace kids (adverse childhood experiences). And I have reason to believe that in less then 10 years time, the present system will come crashing down and be replaced by a system of the future, which will bring in a post scarcity world, make parenting easier, and allow kids to play outside like the way they used to.
This is an important study, and I think the result is real. Earlier access to social media through a personal smartphone appears to predict poorer subsequent performance in maths and Italian. Where I remain cautious is over what we call the cause.
The paper itself cannot tell us why this happens. Its strongest potential mediator is smartphone “pervasiveness” — including use while studying and around sleep — and the authors explicitly raise time displacement and cognitive overload as possible mechanisms.
This is consistent with something I have been exploring in my own work: that sleep displacement may be an important causal pathway behind the international social-media signal. Social media is not a biological mechanism. Sleep disruption is. It affects attention, memory, learning, emotional processing and resilience.
There is also a broader problem here that increasingly concerns me about this debate: selection of evidence.
Jonathan Haidt has done an important job drawing attention to changes in childhood. But evidence consistent with the social-media-harm hypothesis tends to receive considerable attention, while findings that complicate it receive much less.
This paper itself is a good example. The negative maths and Italian results are highlighted. But English shows no negative effect at all, and the authors suggest that digital exposure may actually provide a learning benefit through incidental exposure to English. The strongest mediator is smartphone pervasiveness rather than demonstrated harmful content. And yet the introduction goes further and predicts that repeating the experiment with today's TikTok generation would produce larger effects. That may be true. But it is a hypothesis, not a finding of this study.
This matters because very few parents, teachers or policymakers will read a 16-page Nature Human Behaviour paper and interrogate its methodology. They rely upon people like Haidt — and indeed people like me — to represent the evidence fairly. That places an obligation on all of us to give equal weight to evidence that challenges our preferred hypothesis.
I have been writing on my Substack about this distinction between signal and cause. My reading of the international evidence increasingly leads me towards sleep displacement as a major candidate mechanism. This Italian study doesn't disprove that hypothesis. If anything, its mediation results are remarkably consistent with it.
Social media undoubtedly has costs. It also has benefits. The scientific task is to identify the mechanisms producing each rather than putting everything under a single label.
Before we legislate the signal, we should be much more certain that we have identified the cause.
EDIT: For anyone who is sentient and can spot the irony of using clanker to decide if they can think. This comment illustrates the problem we have: Matthew V. Milone needs to ask AI permission to think. I suggest that you think rather than follow the herd. I notice that you have no comment on the evidence. Maybe use your own mind rather than rely on Pangram for permission to think.
I have similarly looked at his posts, and it is clear that he has a deep misunderstanding of the scientific method.
I agree with you, in a way. It doesn't look egregiously AI-generated, and I'm not sure what raised my initial suspicion to the point of checking Pangram. Regardless, here are a a few points about the comment's style that seem conspicuous to me:
- Leading with cliched, uninformative validation
- Repeated, vague use of the word "signal" (a LinkedIn buzzword) and the comparative lack of medical jargon
- Two sets of emdashes
- It's very punchy. It begins several statements with conjunctions, and it includes several short, declarative sentences.
- It's verbose, yet eloquent.
Furthermore, his other writing, especially his "About" page, does reek of LLM writing. Also, note that he replied to me four times since my allegation, yet none of them dispute it.
Altogether, my guess is that he wrote something himself, perhaps as a list of bullet points, and then used an LLM to connect or polish it.
Goodness me! Talk about fiddling whilst Rome burns. Children's well-being is compromised, and you're behaving like a 6th former who thinks he knows everything. I have a 1986 paper that is apparently AI-assisted. Explain that, or maybe, just maybe, you might shine a light on the message and not the messenger.
I presume you work for Haidt or are looking for a job. Ad hominem attacks are the usual go-to for people who have little idea. I invite you once more to answer: What part of my argument do you disagree with?
These are yet more examples of many hundreds of dubious studies that claim social media and smartphones harm children and teenagers in various ways, from depression to sleeplessness to poor academic performance, suicide, etc., that fail to control for crucial family variables such as violence, psychological abuse, drug/alcohol abuse, mental health problems, criminal behavior, etc., by their parents and household adults. Failing to control for parental/adult abuses and troubles is like studying lung cancer causes while failing to include cigarette smoking.
The CDC's 2021 and 2023 YRB surveys are among the few that do include the key variables of parental abuses and troubles, and these surveys and studies consistently show two major results: compared to teens from healthier families, (1) teens who suffer parental/adult abuses and family troubles are far more likely to have mental health, academic, etc., problems, and (2) teens who are abused and/or have violent, addicted, mentally troubled parents and adults in their homes are much more likely to use social media.
We don't have good measures of family abuse trends, but we do know that drug/alcohol deaths and ER cases rose sharply among parent-age adults during the 2010s, when teens' mental health problems rose. The 2021 and 2023 CDC surveys show 30% of teens grow up with parents and household adults who abuse alcohol/drugs, and one-third with parents who are "severely" mentally troubled and/or violent. Girls and LGBTQ teens report more family abuses and problems, along with poorer mental health.
It is very possible that the simple, usually weak correlation between teens' social media/smartphone use and poorer mental health, suicidality, lack of sleep, cognitive issues, academics, etc., is a false correlation resulting from these problems' connection to family abuses. That issue needs rigorous examination rather than simply being ignored, especially before we rush to ban and restrict teens' social media use.
There's a lot to address in your comment, so I'll focus on what I think is the main point--at least as it applies to this article.
You present an alternative hypothesis, which I'll simplify for the sake of brevity: bad parenting causes both poor academic performance and social media usage. I think there are two reasons why this hypothesis *doesn't* undermine this study's conclusion that social media usage impairs academic achievement.
For one thing, under your hypothesis, the link between abusive parenting and poor academic performance could be *mediated* by social media use. In other words: abusive parents let their kids on social media earlier, and letting kids on social media earlier impairs academic performance. This may be a bit cheeky, but I think it's holding you to the same standard of rigor that you insist on for After Babel and the researchers who they cite.
More substantively, this study *does* control for abusive parenting and other factors because it measures students' post-exposure performance against their pre-exposure performance. If the drop in grades is caused by abuse, then the abuse must begin at the same time that phones are introduced, which would be a suspicious coincidence.
These are very good points and buttress my argument for more rigorous examination of the role of parental/adult abuses and adversities versus social media use in teens' mental health. Regression analysis of the CDC surveys finds these abuses/adversities vastly more important than anything attributable to social media across a wide variety of risks (in particular, social media use is not connected to suicide). Examples of published studies:
Swedo EA et al, Centers for Disease Control. “Adverse childhood experiences and health conditions and risk behaviors among high school students —Youth Risk Behavior Survey, United States, 2023.” MMWR, 73(4);39-49; October 10, 2024. At: https://www.cdc.gov/mmwr/volumes/73/su/su7304a5.htm?s_cid=su7304a5_w
Thériault-Couture F et al. “Cybervictimisation and mental health conditions in young people: findings from a nationally representative longitudinal cohort.” The Lancet Child & Adolescent Health, 10(2);94-102; February 2026. At: https://www.sciencedirect.com/science/article/pii/S2352464225003116
The timing of abuses versus social media usage is important to consider, though symptoms of abuse (ie, suicide) often take years to develop. We should note that the Italian study, like all others on the subject, finds the association between social media acquisitions and academic performance changes is very low (below small to barely small), further indicating the need to include relevant co-founders. We may be risking cutting abused teens off from needed social supports. In any case, you raise good points.
I think you're using a motte-and-bailey in this thread.
Motte: studies about social media's effect on teens should account for parental factors.
Bailey: before we ban social media for teens, we need more studies that account for parental factors.
I agree with the first proposition, and I credit your comment with drawing my attention to that issue. In contrast, I think that the second proposition is very dubious, but I'm hesitant to debate it because of how multifaceted it is, how differently we perceive the literature, and how to frame the issue of social media regulation more broadly.
I'd be willing to debate the bailey given 1) an explicit acknowledgement that the study in the article adequately accounts for parenting, and 2) a commitment to stick to a single aspect of the bailey proposition, such as:
a) the extent to which the research already accounts for parenting
b) the effect sizes of the results
c) the (de)merits of social media restrictions based on uncertain evidence.
Thanks for comment. From its massive 2023 YRB survey, the CDC’s analyses (see previous citation) associate parental/adult abuse, violence, drug/alcohol abuse, and “severe” mental health problems with two-thirds of teens’ poor mental health (sadness), 84% of teens’ opiate abuse, and 89% of teens’ suicide attempts – and frequent social media use with less than 10% of teens’ poor mental health and no significant suicide attempt risk.
That around 21% of girls surveyed report being cyberbullied is deplored. Never mentioned is that 5 in 6 girls who report being cyberbullied ALSO report being violently and/or emotionally abused by parents/grownups. Parental/adult bullying is 3.5 times more prevalent and accounts for nearly all the mental health damage girls report. Analysis of CDC data further shows severely abused girls spend considerably more on social media than non-abused girls.
So, I agree with “bailey” – I don’t think we should continue banning teens from social media until comprehensive analyses clarify these vital factors, which 99% of studies and commentators ignore or control indadequately.
I don't see how this relates to my comment. As I said before, I'm willing to debate the extent to which research already accounts for parenting, but only once we're on the same page about whether the Italian study in particular accounts for it.
"I think there are two reasons why this hypothesis *doesn't* undermine this study's conclusion that social media usage impairs academic achievement" The study does not show that nor cliam that..
Amazing, but we don't have genetically informed data to figure out whether the difference is based upon the conscientiousness of the parent. How much does delaying your kids social media overlap with polygenic risk for educational attainment.
Thanks for clarifying. In that case, my response to you is almost identical to my response to Mile Males:
1. Even if your alternative variable is predictive, its effect on grades might be mediated by social media access.
2. This study implicitly controls for your alternative variable, because the difference in grades do not appear until the children gain social media access.
For point one that would apply if the polygenic score were obtained after the onset of social media, but I think the samples we use are older.
For point two, I get what you are saying, because it's not present in one group and then present in the other. I am wondering how much the difference in early social media access overlaps with the tendency to achieve academically, based upon work that predated social media.
Clinically, I am sure that at least part of the difference is the social media access, but I think that the magnitude of the difference might be overstated by not controlling for genes.
My "skepticism" has been based on dozens of findings that people presumed were causal, and were really just expressions of correlated liabilities to 2 different outcomes.
That framing ignores social media's deleterious effects on objective measures of cognitive abilities--particularly executive functioning.
Even if this was just a difference in style, why would the obligation be on the education system to match social media's style, rather than the other way around?
Studies claiming social media damage to teens' mental health, cognitive functioning, academic achievement, sleeplessness, and other risks fail to control (or inadequately control) the crucial variables of parental/adult abuses and adversities. I can find no exceptions. Meanwhile, those few studies that do control for abuses/adversities find social media use has no effects on mental health, etc. I'm not saying my conclusion is the final word, but we need much better understanding before we risk shutting abused teens out of needed social media contacts.
Ok. I think I see your argument now. I would add that I liked the healthy degree of scepticism you've shown throughout your comments. Here's the thing. The evidence shows that the type of social media matters most. For example, longer-form YouTube videos show little to no signs of negative effects. I think the case could also be made for genuine social media contact.
Unfortunately, this isn't what most kids are consuming. Not even close. The overwhelming majority of the content they are consuming is short viral videos which involve little interactions with others, apart from the ubiquitous mechanisms of like, share and subscribe.
Ofcom's research in the UK showed that 65% of 8–17-year-olds who use social media say they mainly read, watch, follow or like content, compared with 34% who share, comment or post. Once we include for qualitive factors and the fact that people aren't likely to forthcoming about habits which could be perceived by many to be a guilty indulgence, then the figure shrinks even further to support my 'overwhelming majority' framing.
However, your argument also has strong evidence. There is good evidence that social media is an important technological tool in helping kids seek out help when dealing with abuse, especially in terms of accessing peer networks.
I think your strongest argument is a call for nuance. When people lobby for particular legislative changes, the frustration from political inertia can cause them to overlook more nuanced approaches in favour of drastic action. To my mind, a good approach would be a complete ban on the type of short-form viral videos which dominate on TikTok and YouTube shorts, but a Laissez-Faire attitude to everything else.
This would be easy to achieve from a technical viewpoint, and arguments to the contrary are spurious. In December 2019 Mark Ledwich wrote an article on Medium entitled Algorithmic Radicalisation- The Making of a New York Times Myth. It contained an extremely good traffic analysis of YouTube political content which showed that the main algorithmic direction of content steering was towards the corporate centre-Left. If they can steer content, then filtering out harmful, addictive, short-form videos should be an easy matter.
The evidence shows that where there was a bias towards the Right it was user-generated, particularly through people selecting for subject matters which informed them about economics rather than social issues, an area where most people are woefully uninformed or misinformed, especially amongst the young. This also likes explains the growing gender divide in politics. Although it may be controversial to say that males are more interested in things and females are more interested in people, the findings themselves are not at all controversial. Digital media has disabled gatekeeping to the extent that people can now research particular subjects to their own satisfaction.
Your argument is a good one. We shouldn't support blanket bans on social media for the young. Instead we should be far more specific and precise in what we actually want to block. It's easily achieved from an algorithmic perspective, although distinctions about where cut-offs should exist in filtering recommendations is a little more difficult and risks diluting positive change. The real reason why the social media companies have been so reluctant to embrace this type of compromise is because of the attention economy. A Bank of America Global Research estimate put global short-form video (SFV) advertising revenue at about $66bn in 2025. When a product is harmful the market should adapt. It should be noted that the youth market is only a portion of the $66 billion in revenue from SFV.
We don't have precise estimates for how much of global revenue from SFV is youth-orientated, but under-18s were estimated to account for 35% of TikTok's US advertising revenue and 41% of Snapchat's in 2022.
I suspect the reason why we haven't seen a nuanced approach materialise in the political realm is because those politicians who support a ban on social media for the young have political reasons for doing so, whilst those who oppose a ban are equally sure that social media works to their benefit. The real question should be why won't our politicians act like adults?
Good points. We agree that more detailed study including type of social media is needed. It could be that viewers of short-term videos are more troubled to begin with, with short attention spans drawing them to immediacy rather than detail, for example. And yes, there is way too much grownup and corporate self-interest warping the social media debate.
I take your point, but would argue that the same could be said for anything. It’s a bit like the fructose argument, once one gets below the facile veneer. The reason for RFKs reaction to fructose was never because it was substantially less healthy (although it is, but only by a little). The real reason why he took action is because it’s been proven to be more addictive (the rat studies are informative), encouraging more binge behaviour. If one looks at fructose market penetration by country, then the countries with higher rates of fructose penetration tend to have higher consumption, obesity, and higher health costs as a downstream externality caused by the product, particularly in areas like diabetes.
Here’s an interesting titbit. Most sources rate obesity as only moderately polygenic. Robert Plomin, author of Blueprint, takes a different view. He gives it a polygenic score of around 80-90%. That’s because he considers brain mechanisms like reward systems and the ability to self-inhibit an important part of the question of polygenicity, whereas most researchers only look at more obvious physiological factors. His view is that people who feel the strong urge to snack can only tackle the problem by not keeping snacks in the house. It’s worked for him, even though he apparently has a high polygenic score for obesity.
My concerns tend to be the same as yours. I’m a natural libertarian by inclination, and am sceptical giving government any power whatsoever over our daily lives. The exception is children. All the evidence shows that children respond best to high expectations, high discipline cultures in schools. In a recent podcast with Katherine Birbalsingh, headmistress of Britain’s highest performing school, Peter Boghossian was shocked to the core that the British educational establishment wouldn’t want to replicate her successes across the system.
The divide between whether the parent or government should ultimately have the final say over raising children is a somewhat artificial one in this sense. As with most of the better Charters in the US, the opposition to Michaela didn’t come from parents of kids who might attend the school. The opposition came from affluent white women who had the opportunity and the means to send their own children to private schools.
I think the same thing is true of restrictions on social media for children, although there is more nuance. Most of the parents making the loudest noise in opposition are the ones who have the time to stand over their kids shoulder, monitoring their kids online interactions, even though the mechanism is only figurative. That being said I don’t like the nature of the bans as they currently stand. They are too all-encompassing. However, I think most parents would support more selective bans provided it tackled the FOMO issue. Most parents I’ve spoken to hate the fact that their kids are on social media all day, but equally are very wary of their kid missing out if they restrict them from using the technology.
Wow, AI really didn’t like this comment. It wanted me to replace "proven to be more addictive" with "evidence of addiction-like/binge behaviour, particularly in animal models." Bugger it. As usual, I will let the comment stand.
I had an argument on the obesity issue re: polygenic score/heritability. I was a little wrong, but so was the AI. Plomin said his own polygenic risk score for BMI/obesity is at the 94th percentile. However, Plomin and his colleagues specifically examined genetic influence on appetite. In a sample of more than 5,000 twin pairs, they estimated heritability at 63% for satiety responsiveness and 75% for responsiveness to food cues. That’s significantly higher than the standard literature on polygenic risk scores and obesity would suggest, demonstrating a far more complex set of interactions.
Thanks for detail. There is plenty of evidence that what has been touted as discipline, such as school uniforms, zero tolerance, mandatory drug testing, and now cellphone bans are wildly pleasing to educators and parents on an emotional level but don't have long-term benefits. I agree with the high expectations, tho, and am intrigued by the fact that Gen Z girls are demeaned as the most mentally unhealthy but at the same time are the highest-achieving women's generation ever (their mental health "improves" dramatically when they arrive in college).
I already addressed this point under your top-level comment. If you disagree that this Italian study implicitly accounts for other variables, then please go back to that thread to explain why.
The Italian study does not control for parental abuses and adversities, the symptoms of which often are not contemporaneous but can be delayed by years, but only for "parental control practices." Its findings are utterly trivial, with beta values so low that editors should have required authors to state upfront they found next to nothing. It's part of the junk heap of hundreds of poorly-controlled studies whose findings are barely distinguishable from zero.
As I said, this study controls for parental abuses and other factors by looking at the timing. Your hypothesis that there's a delayed effect doesn't alleviate the coincidence. Why would the length of the delay be such that the effect coincides with the introduction of social media, instead of being 2 years too short, 1 year too long, etc.?
Also, I disagree that this effect size is insignificant. As the essay mentions, it's significant by the standards of education literature. To put that effect size into perspective: approximately 60% of students who were prematurely exposed to social media scored worse than the average student who wasn't. Furthermore, if we refined our hypothesis (e.g. to test the effect of specific design features) the effect could be even bigger.
Repeat: this study found nothing. Of course, many tiny effects can become statistically "significant" when sample sizes are large -- ie, eating vegetables is significantly related to teenage depression -- but that is meaningless. The real measure is effect size -- how important is the finding.
Study authors report longitudinal competence score differences between social-media exposed and non-exposed (Beta) of -0.05 sds, -0.06 or -0.04 in some cases. Those translate into effect sizes (Cohen's d values) so abysmally low -- right at nothing-- that what they really mean is that authors found no important relationship between early social media use and school competence scores.
The tinier the effect, the more likely that it would naturally occur during any given period in time. Before-after measures become worthless when effect sizes are near nothing. In fact, some effects during the test period become expected.
I appreciate the opportunity to clarify what these hundreds of apparently "alarming" studies actually are finding. Their authors and advocates have been allowed to imply that 100% of the kids who have cellphones, social media access, etc., suffer cognitive deficiencies (or depression, sleeplessness, etc.), while 100% of those without cellphones etc. are fine.
That is not what these studies find at all. In the Italian case, it found that even at the extremes, teens who got cellphones early and had them for years now suffer cognitive scores less than 1% lower. That is "statistically significant" due to the large sample size, but the effect size (how important cellphone use is to cognition) is close to nothing -- certainly far below "small." Such a minuscule effect size could well be associated far more with parents/grownups who were not abusive, addicted, mentally unwell, jailed, etc., earlier in a child or teen's life but had become more severely troubled later, which the authors did not control for.
In fact, these many hundreds of "alarming" studies, including new ones flooding the media daily, are really finding social media has only trivial effects on children's and teens' mental health, even without including parental abuses. Professionals, academic institutions, and journal editors need to be much more rigorous in demanding that authors include major relevant variables and state they found essentially nothing upfront.
Because of what it is for. I was reminded: "Prepare the kid for the road, not the road for the kid"--Jonathan Haidt. The cognitive environment these kids are in is what the road has become. When every child is orchestrating a horde of AI agents on top of social media interactions, as part of their everyday hourly existence, which is what my life has become (sans social media, don't use it anymore) I don't think modern educational frameworks and practices will be very relevant for measuring flourishing and impact and maybe we're starting to see this mismatch here. I am a dyslexic formerly "retarded" person according to "education", who can now sculpt reality like gods. Just sayin'.
Surely Haidt's quote is not universally true. If food manufacturers put cocaine in their peanut butter, the onus for the educational disruptions that it caused would not be on the schools.
The mediator is the part that stays with me, the pervasiveness rather than the hours. Dinner, sport, the walk home, the dark before sleep. None of that is study time, so none of it looks like learning, and all of it is where a child gets used to staying with something that hasn't started rewarding them yet. Maths punishes a wandering mind more than most subjects, which fits your finding that the gap kept widening there. What thins is the tolerance for an unfinished feeling. That's a hard thing to hand back to someone later.
Im a student. IMO The digital ecosystem by default is incredibly unhealthy right now. U can’t have instagram on your phone without compulsively using it. There needs to be friction between us and engagement-maximizing apps. It is the only way to undo the psychologically manipulative design features. That is why I am such a firm believer in app blockers. They alter my phone ecosystem to undo the addictive design features of certain apps by making them harder to access.
And it is why I believe, for schools, software based phone policy solutions are the future. The phone itself isn’t the problem, it’s the addictive apps within it. put friction between students and the addictive apps so they can learn when to use and not use their phone.
Phones will be in students’ pockets for the rest of their lives, giving them tools to learn self-regulation is the key.
The easiest way to foster positive behavior change is to modify the environment, in this case the digital one. So that it makes being intentional with your phone as easy as possible.
After reading your post and the references offered including the paper 'Longitudinal effect of early social media use on standardized learning outcomes during school year', I resonate with a few of the phrases therein: "AI (has managed) to create stronger engagement for users", "social media use during preadolescence impacts academic outcomes negatively", and the "problem may lie in the continuous fragmentation of attention and the habitual shifting between digital and offline activities". It's no mystery that these "digital distractions" are creating problems for our children. Thanks for your powerful post.
There seems to be an emerging industry of devil's advocates with passionate minions eager to dismiss studies that begin to support what we all know from our observations to be true. Do we really need a rigorous study to explain why people cross a busy intersection, a parking lot, a grocery store aisle without looking up from their phones? Do we need ironclad methodology? Is it really imperative that we eliminate all plausibly exogenous variations in behavior-influencing factors before we conclude that what we are seeing with our own eyes everyday, everywhere is a critically parasitic relationship between the phone and the brain? Do you need that? I don't.
Those charts tell such a clear story. And I find it encouraging that the students who delayed until Grade 9 had better results. That's date we can work with!
Doesn't the English result complicate a pure attention story? Italian and math get nothing back from the feed, but these kids are taking in hours of English music, captions and subtitles from the same platforms, so that null could be two effects cancelling rather than no effect at all.
Very interesting. Personally, I think it is likely to be multiple factors that are responsible for the observed declines, everything from distraction to short form which can only provide so much—unless we evolve into something else, we will realise we cannot package the universe into 5, 10 and 15 second bytes in much the same way as a 2 hour movie often doesn't stand up to the book. Pause, stillness and reflection versus repeat mini-sized, continually replenished gratifications… the acid test might be to compare development to an impossible group—zero social media. A solution. Ensure developmental time outside of SM. Restriction is not working so I imagine a more voluntary or available participation at an early age. I have spend hours of time with my children when they were small catching siri crabs in swampy lagoons on the beach, only to realise that those long hours were just 1 and a hlf to two hours as time had slowed right down, barring a lobotomy, those hours are for life with all their conversation and improvisation. Quality time is a foul expression. Time is everything.
I'm not a social scientist. I'm a middle school teacher. Based on my own anecdotal experience, I would reckon that social media is having a profoundly deleterious cultural effect that is difficult to measure and is collective rather than simply individual (and pscyhological symptomatic). There has never before been a human culture in which the young were allowed to exist in their own ideational spaces in this way, deluged with information and ideas and images and emotions crafted by manipulators and groomers and antisocial actors and profit seekers.
One of the kindergarteners in my school had incest domination porn on his iPad. A 3rd grader had BDSM material. If you're interested in the type of school I teach at you can subscribe to my Substack (or read my most recent post) but that should barely matter. Kids shouldn't be marinating in this content. As a formerly antisocial, impulsive, addiction-prone boy myself I know that this new technological landscape would have ruined my life. I have very little doubt. Where are the adults? As teachers we can't intervene. Parents won't. The bureaucracy (counselors, psychologists, healthcare providers, social workers, etc.) doesn't seem to be too concerned by it all. Perhaps it's just anticipating the wave of dysfunctional clients to come (which means resources and career advance and budget increases and public credibility). Just give it 5-10 years...
https://jmpolemic.substack.com/p/the-maelstrom
I don't think that social media use by kids is going to last. There's more and more restrictions being put up. All you have to do is look at countries like Australia and the UK. If anything, I believe that it's going to get to the point where AI itself starts putting up restrictions, as AI is getting more powerful, even as it's being fed loads of info that social media use by the youth is harmful.
Having said the above, I do share the views of Mike Males, that there's other factors, including negative parenting, that's lowering the mental health of young people. But in a good chunk of these cases, it's the system that's causing negative parenting especially in regards to Ace kids (adverse childhood experiences). And I have reason to believe that in less then 10 years time, the present system will come crashing down and be replaced by a system of the future, which will bring in a post scarcity world, make parenting easier, and allow kids to play outside like the way they used to.
Ultimately it’s upon the family to restrict access and passwords.
This is an important study, and I think the result is real. Earlier access to social media through a personal smartphone appears to predict poorer subsequent performance in maths and Italian. Where I remain cautious is over what we call the cause.
The paper itself cannot tell us why this happens. Its strongest potential mediator is smartphone “pervasiveness” — including use while studying and around sleep — and the authors explicitly raise time displacement and cognitive overload as possible mechanisms.
This is consistent with something I have been exploring in my own work: that sleep displacement may be an important causal pathway behind the international social-media signal. Social media is not a biological mechanism. Sleep disruption is. It affects attention, memory, learning, emotional processing and resilience.
There is also a broader problem here that increasingly concerns me about this debate: selection of evidence.
Jonathan Haidt has done an important job drawing attention to changes in childhood. But evidence consistent with the social-media-harm hypothesis tends to receive considerable attention, while findings that complicate it receive much less.
This paper itself is a good example. The negative maths and Italian results are highlighted. But English shows no negative effect at all, and the authors suggest that digital exposure may actually provide a learning benefit through incidental exposure to English. The strongest mediator is smartphone pervasiveness rather than demonstrated harmful content. And yet the introduction goes further and predicts that repeating the experiment with today's TikTok generation would produce larger effects. That may be true. But it is a hypothesis, not a finding of this study.
This matters because very few parents, teachers or policymakers will read a 16-page Nature Human Behaviour paper and interrogate its methodology. They rely upon people like Haidt — and indeed people like me — to represent the evidence fairly. That places an obligation on all of us to give equal weight to evidence that challenges our preferred hypothesis.
I have been writing on my Substack about this distinction between signal and cause. My reading of the international evidence increasingly leads me towards sleep displacement as a major candidate mechanism. This Italian study doesn't disprove that hypothesis. If anything, its mediation results are remarkably consistent with it.
Social media undoubtedly has costs. It also has benefits. The scientific task is to identify the mechanisms producing each rather than putting everything under a single label.
Before we legislate the signal, we should be much more certain that we have identified the cause.
Begone, clanker.
EDIT: For anyone who's curious, I checked this comment on Pangram and manually spot-checked his other recent comments before making this accusation.
EDIT: For anyone who is sentient and can spot the irony of using clanker to decide if they can think. This comment illustrates the problem we have: Matthew V. Milone needs to ask AI permission to think. I suggest that you think rather than follow the herd. I notice that you have no comment on the evidence. Maybe use your own mind rather than rely on Pangram for permission to think.
I have similarly looked at his posts, and it is clear that he has a deep misunderstanding of the scientific method.
This does not read like AI to me at all. Pangram has known false positives.
Yes, and humans have known false negatives. ;)
I agree with you, in a way. It doesn't look egregiously AI-generated, and I'm not sure what raised my initial suspicion to the point of checking Pangram. Regardless, here are a a few points about the comment's style that seem conspicuous to me:
- Leading with cliched, uninformative validation
- Repeated, vague use of the word "signal" (a LinkedIn buzzword) and the comparative lack of medical jargon
- Two sets of emdashes
- It's very punchy. It begins several statements with conjunctions, and it includes several short, declarative sentences.
- It's verbose, yet eloquent.
Furthermore, his other writing, especially his "About" page, does reek of LLM writing. Also, note that he replied to me four times since my allegation, yet none of them dispute it.
Altogether, my guess is that he wrote something himself, perhaps as a list of bullet points, and then used an LLM to connect or polish it.
Goodness me! Talk about fiddling whilst Rome burns. Children's well-being is compromised, and you're behaving like a 6th former who thinks he knows everything. I have a 1986 paper that is apparently AI-assisted. Explain that, or maybe, just maybe, you might shine a light on the message and not the messenger.
I presume you work for Haidt or are looking for a job. Ad hominem attacks are the usual go-to for people who have little idea. I invite you once more to answer: What part of my argument do you disagree with?
Happy to discuss the evidence. Which part of the argument do you think is wrong?
If the comment wasn't worth your time to write, then it probably isn't worth my time to read.
https://www.lesswrong.com/posts/kjQdL3dxaACSbjkSx/why-you-should-almost-never-use-ai-to-write-anything-1
These are yet more examples of many hundreds of dubious studies that claim social media and smartphones harm children and teenagers in various ways, from depression to sleeplessness to poor academic performance, suicide, etc., that fail to control for crucial family variables such as violence, psychological abuse, drug/alcohol abuse, mental health problems, criminal behavior, etc., by their parents and household adults. Failing to control for parental/adult abuses and troubles is like studying lung cancer causes while failing to include cigarette smoking.
The CDC's 2021 and 2023 YRB surveys are among the few that do include the key variables of parental abuses and troubles, and these surveys and studies consistently show two major results: compared to teens from healthier families, (1) teens who suffer parental/adult abuses and family troubles are far more likely to have mental health, academic, etc., problems, and (2) teens who are abused and/or have violent, addicted, mentally troubled parents and adults in their homes are much more likely to use social media.
We don't have good measures of family abuse trends, but we do know that drug/alcohol deaths and ER cases rose sharply among parent-age adults during the 2010s, when teens' mental health problems rose. The 2021 and 2023 CDC surveys show 30% of teens grow up with parents and household adults who abuse alcohol/drugs, and one-third with parents who are "severely" mentally troubled and/or violent. Girls and LGBTQ teens report more family abuses and problems, along with poorer mental health.
It is very possible that the simple, usually weak correlation between teens' social media/smartphone use and poorer mental health, suicidality, lack of sleep, cognitive issues, academics, etc., is a false correlation resulting from these problems' connection to family abuses. That issue needs rigorous examination rather than simply being ignored, especially before we rush to ban and restrict teens' social media use.
There's a lot to address in your comment, so I'll focus on what I think is the main point--at least as it applies to this article.
You present an alternative hypothesis, which I'll simplify for the sake of brevity: bad parenting causes both poor academic performance and social media usage. I think there are two reasons why this hypothesis *doesn't* undermine this study's conclusion that social media usage impairs academic achievement.
For one thing, under your hypothesis, the link between abusive parenting and poor academic performance could be *mediated* by social media use. In other words: abusive parents let their kids on social media earlier, and letting kids on social media earlier impairs academic performance. This may be a bit cheeky, but I think it's holding you to the same standard of rigor that you insist on for After Babel and the researchers who they cite.
More substantively, this study *does* control for abusive parenting and other factors because it measures students' post-exposure performance against their pre-exposure performance. If the drop in grades is caused by abuse, then the abuse must begin at the same time that phones are introduced, which would be a suspicious coincidence.
These are very good points and buttress my argument for more rigorous examination of the role of parental/adult abuses and adversities versus social media use in teens' mental health. Regression analysis of the CDC surveys finds these abuses/adversities vastly more important than anything attributable to social media across a wide variety of risks (in particular, social media use is not connected to suicide). Examples of published studies:
Swedo EA et al, Centers for Disease Control. “Adverse childhood experiences and health conditions and risk behaviors among high school students —Youth Risk Behavior Survey, United States, 2023.” MMWR, 73(4);39-49; October 10, 2024. At: https://www.cdc.gov/mmwr/volumes/73/su/su7304a5.htm?s_cid=su7304a5_w
Thériault-Couture F et al. “Cybervictimisation and mental health conditions in young people: findings from a nationally representative longitudinal cohort.” The Lancet Child & Adolescent Health, 10(2);94-102; February 2026. At: https://www.sciencedirect.com/science/article/pii/S2352464225003116
Ferguson CJ. “Links between social media use and mental wellness in youth are an artifact of other factors: implications for public policy and meta analysis.” Current Psychology 45:573; 9 February 2026. At: https://www.christopherjferguson.com/Brainwaves.pdf?utm_source=substack&utm_medium=email
Ferguson CJ. “Adverse childhood events, not age of acquiring smartphones or tablets, predict mental health in young adults.” Clinical Neuropsychiatry, 23(3); 243-247; 2026. At: https://www.clinicalneuropsychiatry.org/download/adverse-childhood-events-not-age-of-acquiring-smartphones-or-tablets-predict-mental-health-in-young-adults/
Vuorre M et al. “Global well-being and mental health in the internet age.” Clinical Psychological Science, 12(5);November 27, 2023. At: https://journals.sagepub.com/doi/10.1177/21677026231207791
The timing of abuses versus social media usage is important to consider, though symptoms of abuse (ie, suicide) often take years to develop. We should note that the Italian study, like all others on the subject, finds the association between social media acquisitions and academic performance changes is very low (below small to barely small), further indicating the need to include relevant co-founders. We may be risking cutting abused teens off from needed social supports. In any case, you raise good points.
I think you're using a motte-and-bailey in this thread.
Motte: studies about social media's effect on teens should account for parental factors.
Bailey: before we ban social media for teens, we need more studies that account for parental factors.
I agree with the first proposition, and I credit your comment with drawing my attention to that issue. In contrast, I think that the second proposition is very dubious, but I'm hesitant to debate it because of how multifaceted it is, how differently we perceive the literature, and how to frame the issue of social media regulation more broadly.
I'd be willing to debate the bailey given 1) an explicit acknowledgement that the study in the article adequately accounts for parenting, and 2) a commitment to stick to a single aspect of the bailey proposition, such as:
a) the extent to which the research already accounts for parenting
b) the effect sizes of the results
c) the (de)merits of social media restrictions based on uncertain evidence.
You can pick the category.
Thanks for comment. From its massive 2023 YRB survey, the CDC’s analyses (see previous citation) associate parental/adult abuse, violence, drug/alcohol abuse, and “severe” mental health problems with two-thirds of teens’ poor mental health (sadness), 84% of teens’ opiate abuse, and 89% of teens’ suicide attempts – and frequent social media use with less than 10% of teens’ poor mental health and no significant suicide attempt risk.
That around 21% of girls surveyed report being cyberbullied is deplored. Never mentioned is that 5 in 6 girls who report being cyberbullied ALSO report being violently and/or emotionally abused by parents/grownups. Parental/adult bullying is 3.5 times more prevalent and accounts for nearly all the mental health damage girls report. Analysis of CDC data further shows severely abused girls spend considerably more on social media than non-abused girls.
So, I agree with “bailey” – I don’t think we should continue banning teens from social media until comprehensive analyses clarify these vital factors, which 99% of studies and commentators ignore or control indadequately.
I don't see how this relates to my comment. As I said before, I'm willing to debate the extent to which research already accounts for parenting, but only once we're on the same page about whether the Italian study in particular accounts for it.
"I think there are two reasons why this hypothesis *doesn't* undermine this study's conclusion that social media usage impairs academic achievement" The study does not show that nor cliam that..
Amazing, but we don't have genetically informed data to figure out whether the difference is based upon the conscientiousness of the parent. How much does delaying your kids social media overlap with polygenic risk for educational attainment.
Is this satire?
No, it's a solid scientific question. Correlation versus causality. We gave a tool to assess it.
Thanks for clarifying. In that case, my response to you is almost identical to my response to Mile Males:
1. Even if your alternative variable is predictive, its effect on grades might be mediated by social media access.
2. This study implicitly controls for your alternative variable, because the difference in grades do not appear until the children gain social media access.
For point one that would apply if the polygenic score were obtained after the onset of social media, but I think the samples we use are older.
For point two, I get what you are saying, because it's not present in one group and then present in the other. I am wondering how much the difference in early social media access overlaps with the tendency to achieve academically, based upon work that predated social media.
Clinically, I am sure that at least part of the difference is the social media access, but I think that the magnitude of the difference might be overstated by not controlling for genes.
My "skepticism" has been based on dozens of findings that people presumed were causal, and were really just expressions of correlated liabilities to 2 different outcomes.
I think you could reframe this whole debate as: modern education fails to catch up with Gen Z's smartphone-influenced cognitive styles.
That framing ignores social media's deleterious effects on objective measures of cognitive abilities--particularly executive functioning.
Even if this was just a difference in style, why would the obligation be on the education system to match social media's style, rather than the other way around?
Studies claiming social media damage to teens' mental health, cognitive functioning, academic achievement, sleeplessness, and other risks fail to control (or inadequately control) the crucial variables of parental/adult abuses and adversities. I can find no exceptions. Meanwhile, those few studies that do control for abuses/adversities find social media use has no effects on mental health, etc. I'm not saying my conclusion is the final word, but we need much better understanding before we risk shutting abused teens out of needed social media contacts.
Ok. I think I see your argument now. I would add that I liked the healthy degree of scepticism you've shown throughout your comments. Here's the thing. The evidence shows that the type of social media matters most. For example, longer-form YouTube videos show little to no signs of negative effects. I think the case could also be made for genuine social media contact.
Unfortunately, this isn't what most kids are consuming. Not even close. The overwhelming majority of the content they are consuming is short viral videos which involve little interactions with others, apart from the ubiquitous mechanisms of like, share and subscribe.
Ofcom's research in the UK showed that 65% of 8–17-year-olds who use social media say they mainly read, watch, follow or like content, compared with 34% who share, comment or post. Once we include for qualitive factors and the fact that people aren't likely to forthcoming about habits which could be perceived by many to be a guilty indulgence, then the figure shrinks even further to support my 'overwhelming majority' framing.
However, your argument also has strong evidence. There is good evidence that social media is an important technological tool in helping kids seek out help when dealing with abuse, especially in terms of accessing peer networks.
I think your strongest argument is a call for nuance. When people lobby for particular legislative changes, the frustration from political inertia can cause them to overlook more nuanced approaches in favour of drastic action. To my mind, a good approach would be a complete ban on the type of short-form viral videos which dominate on TikTok and YouTube shorts, but a Laissez-Faire attitude to everything else.
This would be easy to achieve from a technical viewpoint, and arguments to the contrary are spurious. In December 2019 Mark Ledwich wrote an article on Medium entitled Algorithmic Radicalisation- The Making of a New York Times Myth. It contained an extremely good traffic analysis of YouTube political content which showed that the main algorithmic direction of content steering was towards the corporate centre-Left. If they can steer content, then filtering out harmful, addictive, short-form videos should be an easy matter.
The evidence shows that where there was a bias towards the Right it was user-generated, particularly through people selecting for subject matters which informed them about economics rather than social issues, an area where most people are woefully uninformed or misinformed, especially amongst the young. This also likes explains the growing gender divide in politics. Although it may be controversial to say that males are more interested in things and females are more interested in people, the findings themselves are not at all controversial. Digital media has disabled gatekeeping to the extent that people can now research particular subjects to their own satisfaction.
Your argument is a good one. We shouldn't support blanket bans on social media for the young. Instead we should be far more specific and precise in what we actually want to block. It's easily achieved from an algorithmic perspective, although distinctions about where cut-offs should exist in filtering recommendations is a little more difficult and risks diluting positive change. The real reason why the social media companies have been so reluctant to embrace this type of compromise is because of the attention economy. A Bank of America Global Research estimate put global short-form video (SFV) advertising revenue at about $66bn in 2025. When a product is harmful the market should adapt. It should be noted that the youth market is only a portion of the $66 billion in revenue from SFV.
We don't have precise estimates for how much of global revenue from SFV is youth-orientated, but under-18s were estimated to account for 35% of TikTok's US advertising revenue and 41% of Snapchat's in 2022.
I suspect the reason why we haven't seen a nuanced approach materialise in the political realm is because those politicians who support a ban on social media for the young have political reasons for doing so, whilst those who oppose a ban are equally sure that social media works to their benefit. The real question should be why won't our politicians act like adults?
Good points. We agree that more detailed study including type of social media is needed. It could be that viewers of short-term videos are more troubled to begin with, with short attention spans drawing them to immediacy rather than detail, for example. And yes, there is way too much grownup and corporate self-interest warping the social media debate.
I take your point, but would argue that the same could be said for anything. It’s a bit like the fructose argument, once one gets below the facile veneer. The reason for RFKs reaction to fructose was never because it was substantially less healthy (although it is, but only by a little). The real reason why he took action is because it’s been proven to be more addictive (the rat studies are informative), encouraging more binge behaviour. If one looks at fructose market penetration by country, then the countries with higher rates of fructose penetration tend to have higher consumption, obesity, and higher health costs as a downstream externality caused by the product, particularly in areas like diabetes.
Here’s an interesting titbit. Most sources rate obesity as only moderately polygenic. Robert Plomin, author of Blueprint, takes a different view. He gives it a polygenic score of around 80-90%. That’s because he considers brain mechanisms like reward systems and the ability to self-inhibit an important part of the question of polygenicity, whereas most researchers only look at more obvious physiological factors. His view is that people who feel the strong urge to snack can only tackle the problem by not keeping snacks in the house. It’s worked for him, even though he apparently has a high polygenic score for obesity.
My concerns tend to be the same as yours. I’m a natural libertarian by inclination, and am sceptical giving government any power whatsoever over our daily lives. The exception is children. All the evidence shows that children respond best to high expectations, high discipline cultures in schools. In a recent podcast with Katherine Birbalsingh, headmistress of Britain’s highest performing school, Peter Boghossian was shocked to the core that the British educational establishment wouldn’t want to replicate her successes across the system.
The divide between whether the parent or government should ultimately have the final say over raising children is a somewhat artificial one in this sense. As with most of the better Charters in the US, the opposition to Michaela didn’t come from parents of kids who might attend the school. The opposition came from affluent white women who had the opportunity and the means to send their own children to private schools.
I think the same thing is true of restrictions on social media for children, although there is more nuance. Most of the parents making the loudest noise in opposition are the ones who have the time to stand over their kids shoulder, monitoring their kids online interactions, even though the mechanism is only figurative. That being said I don’t like the nature of the bans as they currently stand. They are too all-encompassing. However, I think most parents would support more selective bans provided it tackled the FOMO issue. Most parents I’ve spoken to hate the fact that their kids are on social media all day, but equally are very wary of their kid missing out if they restrict them from using the technology.
Wow, AI really didn’t like this comment. It wanted me to replace "proven to be more addictive" with "evidence of addiction-like/binge behaviour, particularly in animal models." Bugger it. As usual, I will let the comment stand.
I had an argument on the obesity issue re: polygenic score/heritability. I was a little wrong, but so was the AI. Plomin said his own polygenic risk score for BMI/obesity is at the 94th percentile. However, Plomin and his colleagues specifically examined genetic influence on appetite. In a sample of more than 5,000 twin pairs, they estimated heritability at 63% for satiety responsiveness and 75% for responsiveness to food cues. That’s significantly higher than the standard literature on polygenic risk scores and obesity would suggest, demonstrating a far more complex set of interactions.
Thanks for detail. There is plenty of evidence that what has been touted as discipline, such as school uniforms, zero tolerance, mandatory drug testing, and now cellphone bans are wildly pleasing to educators and parents on an emotional level but don't have long-term benefits. I agree with the high expectations, tho, and am intrigued by the fact that Gen Z girls are demeaned as the most mentally unhealthy but at the same time are the highest-achieving women's generation ever (their mental health "improves" dramatically when they arrive in college).
I already addressed this point under your top-level comment. If you disagree that this Italian study implicitly accounts for other variables, then please go back to that thread to explain why.
The Italian study does not control for parental abuses and adversities, the symptoms of which often are not contemporaneous but can be delayed by years, but only for "parental control practices." Its findings are utterly trivial, with beta values so low that editors should have required authors to state upfront they found next to nothing. It's part of the junk heap of hundreds of poorly-controlled studies whose findings are barely distinguishable from zero.
As I said, this study controls for parental abuses and other factors by looking at the timing. Your hypothesis that there's a delayed effect doesn't alleviate the coincidence. Why would the length of the delay be such that the effect coincides with the introduction of social media, instead of being 2 years too short, 1 year too long, etc.?
Also, I disagree that this effect size is insignificant. As the essay mentions, it's significant by the standards of education literature. To put that effect size into perspective: approximately 60% of students who were prematurely exposed to social media scored worse than the average student who wasn't. Furthermore, if we refined our hypothesis (e.g. to test the effect of specific design features) the effect could be even bigger.
Repeat: this study found nothing. Of course, many tiny effects can become statistically "significant" when sample sizes are large -- ie, eating vegetables is significantly related to teenage depression -- but that is meaningless. The real measure is effect size -- how important is the finding.
Study authors report longitudinal competence score differences between social-media exposed and non-exposed (Beta) of -0.05 sds, -0.06 or -0.04 in some cases. Those translate into effect sizes (Cohen's d values) so abysmally low -- right at nothing-- that what they really mean is that authors found no important relationship between early social media use and school competence scores.
The tinier the effect, the more likely that it would naturally occur during any given period in time. Before-after measures become worthless when effect sizes are near nothing. In fact, some effects during the test period become expected.
I appreciate the opportunity to clarify what these hundreds of apparently "alarming" studies actually are finding. Their authors and advocates have been allowed to imply that 100% of the kids who have cellphones, social media access, etc., suffer cognitive deficiencies (or depression, sleeplessness, etc.), while 100% of those without cellphones etc. are fine.
That is not what these studies find at all. In the Italian case, it found that even at the extremes, teens who got cellphones early and had them for years now suffer cognitive scores less than 1% lower. That is "statistically significant" due to the large sample size, but the effect size (how important cellphone use is to cognition) is close to nothing -- certainly far below "small." Such a minuscule effect size could well be associated far more with parents/grownups who were not abusive, addicted, mentally unwell, jailed, etc., earlier in a child or teen's life but had become more severely troubled later, which the authors did not control for.
In fact, these many hundreds of "alarming" studies, including new ones flooding the media daily, are really finding social media has only trivial effects on children's and teens' mental health, even without including parental abuses. Professionals, academic institutions, and journal editors need to be much more rigorous in demanding that authors include major relevant variables and state they found essentially nothing upfront.
Because of what it is for. I was reminded: "Prepare the kid for the road, not the road for the kid"--Jonathan Haidt. The cognitive environment these kids are in is what the road has become. When every child is orchestrating a horde of AI agents on top of social media interactions, as part of their everyday hourly existence, which is what my life has become (sans social media, don't use it anymore) I don't think modern educational frameworks and practices will be very relevant for measuring flourishing and impact and maybe we're starting to see this mismatch here. I am a dyslexic formerly "retarded" person according to "education", who can now sculpt reality like gods. Just sayin'.
Surely Haidt's quote is not universally true. If food manufacturers put cocaine in their peanut butter, the onus for the educational disruptions that it caused would not be on the schools.
The mediator is the part that stays with me, the pervasiveness rather than the hours. Dinner, sport, the walk home, the dark before sleep. None of that is study time, so none of it looks like learning, and all of it is where a child gets used to staying with something that hasn't started rewarding them yet. Maths punishes a wandering mind more than most subjects, which fits your finding that the gap kept widening there. What thins is the tolerance for an unfinished feeling. That's a hard thing to hand back to someone later.
Im a student. IMO The digital ecosystem by default is incredibly unhealthy right now. U can’t have instagram on your phone without compulsively using it. There needs to be friction between us and engagement-maximizing apps. It is the only way to undo the psychologically manipulative design features. That is why I am such a firm believer in app blockers. They alter my phone ecosystem to undo the addictive design features of certain apps by making them harder to access.
And it is why I believe, for schools, software based phone policy solutions are the future. The phone itself isn’t the problem, it’s the addictive apps within it. put friction between students and the addictive apps so they can learn when to use and not use their phone.
Phones will be in students’ pockets for the rest of their lives, giving them tools to learn self-regulation is the key.
The easiest way to foster positive behavior change is to modify the environment, in this case the digital one. So that it makes being intentional with your phone as easy as possible.
After reading your post and the references offered including the paper 'Longitudinal effect of early social media use on standardized learning outcomes during school year', I resonate with a few of the phrases therein: "AI (has managed) to create stronger engagement for users", "social media use during preadolescence impacts academic outcomes negatively", and the "problem may lie in the continuous fragmentation of attention and the habitual shifting between digital and offline activities". It's no mystery that these "digital distractions" are creating problems for our children. Thanks for your powerful post.
Nice to have data about this.
Our kids can get social media and the smartphone at eighteen
We also took away their tablets. They don't get underestricted access to the laptop either
It's in the family room and password protected
There seems to be an emerging industry of devil's advocates with passionate minions eager to dismiss studies that begin to support what we all know from our observations to be true. Do we really need a rigorous study to explain why people cross a busy intersection, a parking lot, a grocery store aisle without looking up from their phones? Do we need ironclad methodology? Is it really imperative that we eliminate all plausibly exogenous variations in behavior-influencing factors before we conclude that what we are seeing with our own eyes everyday, everywhere is a critically parasitic relationship between the phone and the brain? Do you need that? I don't.
The real question is what early access changes over time, not just what happens in the moment.
Those charts tell such a clear story. And I find it encouraging that the students who delayed until Grade 9 had better results. That's date we can work with!
Saved and look forward to reading this article.
Doesn't the English result complicate a pure attention story? Italian and math get nothing back from the feed, but these kids are taking in hours of English music, captions and subtitles from the same platforms, so that null could be two effects cancelling rather than no effect at all.
Very interesting. Personally, I think it is likely to be multiple factors that are responsible for the observed declines, everything from distraction to short form which can only provide so much—unless we evolve into something else, we will realise we cannot package the universe into 5, 10 and 15 second bytes in much the same way as a 2 hour movie often doesn't stand up to the book. Pause, stillness and reflection versus repeat mini-sized, continually replenished gratifications… the acid test might be to compare development to an impossible group—zero social media. A solution. Ensure developmental time outside of SM. Restriction is not working so I imagine a more voluntary or available participation at an early age. I have spend hours of time with my children when they were small catching siri crabs in swampy lagoons on the beach, only to realise that those long hours were just 1 and a hlf to two hours as time had slowed right down, barring a lobotomy, those hours are for life with all their conversation and improvisation. Quality time is a foul expression. Time is everything.