1. Haid has evidence from randomized experiments with kids.
2. There are natural experiments caused by different regional adoption rates of high-speed internet.
3. When the effect size of harm correlates with the frequency of use in population, that's more than time series and spurious-correlations.
Social Media and Mental Health: A Collaborative Review https://docs.google.com/document/d/1w-HOfseF2wF9YIpXwUUtP65-...
https://www.thedailybeast.com/the-coddling-of-the-american-p...
One of the points he makes -- and that I don't see much discussed when it comes to Haidt's work -- is that Haidt's prescriptions for a solution are dangerous. If Haidt is wrong, one of the most unsettling consequences of relying on his solutions is that we stop looking for the real causes of any degradation in childhood mental health. Even if he's right, his solutions will have all sorts of unintended consequences, including loss of privacy on the internet, lack of access by children to resources that are in fact helpful to their mental health, and violations of various legal and constitutional human rights frameworks.
I'm not sure how to put this, but effect size and causation are different things. It's a category mistake to compare.
It's not necessary. Randomized experiment is good enough. So are natural experiments.