Endogenous factors are addressed in the paper. Nothing personal, but I'm getting so tired of people on HN saying 'well, maybe they made an elementary statistical mistake...' They're professional statisticans working with a large and robust dataset,a nd they devoted 6-7 pages of the paper to discussing how they controlled for such factors. And they brought up your exact point in the very first paragraph of that discussion:
[...] In particular, while the effect on well-being has been largely explored in the case of face-to-face interactions, we do not have conclusive evidence about the endogenous relationship between online interactions and well-being. This weakness suggests caution about the generality of the results provided by previous literature. Individual effects such as personal characteristics may be correlated with both participation in SNSs and well-being. Happier people may also be more outgoing and open-minded, and may have a higher propensity for various kinds of social interaction. [...]
It's fine to be skeptical, but couldn't you at least do the authors the courtesy of looking at their methods and pointing out where you think they fall short?
I agree that we should always at least read the linked article (and sometimes the original paper) before bringing out "correlation does not equal causation".
However, you are overstating the case for the author's having dealt with reverse causality. From the conclusion:
When we addressed causality in IV estimates, the significance of the correlation between participation in social networking sites and subjective well-being disappeared.
See my top level comment for my own take on the article.
I'm not trying to make any comment on the significance of the result; the authors discuss those factors in detail in the body of the paper (around page 20) as well as mentioning the limitations in the abstract and conclusion.
My gripe is with reflexive and unjustified statistical one-upmanship in HN discussions.
[...] In particular, while the effect on well-being has been largely explored in the case of face-to-face interactions, we do not have conclusive evidence about the endogenous relationship between online interactions and well-being. This weakness suggests caution about the generality of the results provided by previous literature. Individual effects such as personal characteristics may be correlated with both participation in SNSs and well-being. Happier people may also be more outgoing and open-minded, and may have a higher propensity for various kinds of social interaction. [...]
It's fine to be skeptical, but couldn't you at least do the authors the courtesy of looking at their methods and pointing out where you think they fall short?