I don't see how this blog article contains any scientific rigor by not doing any on-the-ground surveys in all these countries, but instead relies upon analytics from their own platform.
But I definitely believe there are too many factors within the data for a full understanding. Most likely a case of Correlation != Causation