They count public posts. Now that G+ has communities and they are all by default public postings, I can see G+ jumping up the rankings. All the previous studies that talked about how G+ was struggling also looked at public postings only. I've been using G+ since it was still invite only. I and the hundreds of folks in my circles were not posting publicly so as far as anyone could tell my account was unused. Now I have a lot of public posts as I'm very active in my communities.
Twitter and Facebook are public postings by default so it has been easier for these third parties to try to determine how active the user base is at those sites.