I'm not a parent and hardly use social media (well, except HN!), but I am still willing to deny this, because Haidt has never properly addressed the serious demolition of his work presented by Aaron Brown. Have we all forgotten how weak and FP prone social studies are?
To Haidt's credit he did at least try a rebuttal, which is better than most situations in modern academia where outside criticisms are just ignored entirely. But he lost that debate because his responses were bad, and in this article he doubles down on the bad arguments he made last time.
He attempts to suggest that the large number of studies he collected with Twenge means there must be something there. This argument fails because when Brown randomly spot checked some of his cited studies they were all of garbage quality. Often they had nothing to do with social media at all. Haidt's response was the same as in this article: they can't all be wrong! But yes, yes they can all be wrong and Brown has strongly shown that this is likely to be the case.
Then Haidt suggests that you can derive signal from a large pile of bad studies, because if the null hypothesis were true then you'd see random results. But you can't assume that due to publishing biases, spurious correlations and other problems. For example he says that if social media didn't affect mental health, then there'd be no gender signal. That doesn't make sense. It's possible for girls to have worse mental health, and spend more time on social media, and for social media to not be the cause.
After making spurious correlation/causation arguments and denying he's doing so, he moves on to an even worse problem making his entire theory is unfalsifiable:
> much of my book is about the collective action traps that entire communities of adolescents fall into when they move their social lives onto these platforms, such that it becomes costly to abstain. It is at that point that collective mental health declines most sharply, and the individuals who try to quit find that they are socially isolated. The skeptics do not consider the ways that these network or group-level effects may obscure individual-level effects, and may be much larger than the individual-level effects.
In other words you can't even attempt to establish causality by cutting out social media use and seeing if it affects depression, because he will just immediately change his claim to assert that it was the ambient social-media-ness of society causing the depression (a much harder hypothesis to prove), and thus individual interventions can't help. And because this hypothesis is so vague, he can always fall back to claiming that whatever group intervention size is tried it wasn't big enough.
His whole essay is full of problems like this. It's basically a Gish Gallop. Vast reams of unusable paper is presented, along with panicked assertions that it's unreasonable to request valid evidence or a testable hypothesis before making law, due to the nature of the (self declared) emergency: a classic circular argument of the sort that emerges way too often in public health spheres.
There's actually a better supported alternative line of work that shows a completely different and much stronger correlation related to political ideology. But of course Haidt would rather not discuss that, as it's not amenable to legislation.