One can always object that correlation doesn't equal causation. They talk in the article about controlling for certain variables as best one can. You don't get

One can always object that correlation doesn't equal causation. They talk in the article about controlling for certain variables as best one can. You don't get to run double-blind experiments on societies, but as they say in the article, this is about as close as one gets. Your idea about the web is certainly interesting, and could be explored further, but it doesn't strongly account for why a specific age group is effected and why areas with higher abortion access are more effected while low access areas are less effected. Etc. Etc. Taken as a whole, the data do seem to make a strong case for their theory.