A lot of the stats cited are ultimately subjective assessments of things like mental health. Calling more people "bipolar" or "depressed" or "anxious" can just as easily be a change in language as to what those words mean, as easily as they can reflect a change in the underlying language and how people express things. Notably the definition of most mental illnesses in the country Haidt is pulling his data from changed in 2013 due to the release of the DSM-V, so I don't really think pre-2013 data is comparable to post-2013 data at all, as these words in question are not medically or socially defined in the same way that they were before. In fact one of the most common criticisms of the DSM-V when it was being drafted to the present day is the allegation that it leads to more mental illness diagnosis's.

Attempted suicide is one of the few stats I treat as grounded relatively firmly in reality because it's not nearly as subjective, but this measure seems to have started skyrocketing in the mid 2000's not in 2012.

If the conclusion is that the rise of the internet has caused more and more teens to label themselves as mentally ill, I would say that's a conclusion I'm firmly convinced of. The conclusion that mental illness is actually increasing population wide, I'm very very sceptical of. I could imagine that it's happening to an extent due to things like economic pressures and increased inequality, we are seeing things like declines in lifespans and increases in suicide, but I don't think mental illness is increasingly nearly as much as the stats would lead you to believe at face value.

Plus quite simply, using time series data about when mental illness increased in society doesn't really shed much light on WHAT caused the mental illness to rise even if you can establish such a rise was happening, but I'm not even convinced by the date "2012".