This is my summary based on skimming the article: The authors find that social network usage has a negative correlation with happiness. To control for unhappy

This is my summary based on skimming the article: The authors find that social network usage has a negative correlation with happiness. To control for unhappy people turning to social networks (reverse causality), the use instrumental variables (IV). This involves finding an instrument, that is something that causes people to use social networks more/less in a quasi-random way. Their instrument is availability of DSL. I'm not sure how they deal with DSL availability also affecting happiness through other mechanisms (e.g. general internet usage). In any case, the correlation between social network usage and happiness is not significant once they use the IV strategy. They then use a method I'm not familiar with (structural equation modelling, SEM) to obtain some other results which are discussed in the abstract. I believe that the abstract (and the popular article discussing it) should have mentioned the insignificance of the correlation when using the IV strategy, since I doubt many casual readers would by able to critically evaluate the SEM strategy (which yielded the results that the abstract and journalistic article discussed).