It seems a bit naive to say "every story in the world" when they analyzed only "thousands of novels" -- this doesn't even mention what languages they analyzed. I could imagine many Greek-influenced cultures (e.g. "the West") having similar archetypes by virtue of a common ancestor. Language analysis tools for CJK aren't, as far as I know, as advanced, so I can imagine a lot of stories from those languages being left out as well.
The reason I bring this up is that I'm reading a book called "Don't Sleep, There Are Snakes,"  which is about a remote tribe of people in the Amazon jungle called the Pirahã. 
One of the most stunning points in the book is that their culture is such that all stories told by native Pirahã are based on first-person experience. When a Pirahã dies, their stories are not passed on or retold. Due to this, there's also no need to record past stories, orally or otherwise.
This kind of study would obviously exclude stories from that kind of culture and seems a bit narrow-minded: "everything I looked at says A, so EVERYTHING must be A."