Another approach (which would be time-consuming, admittedly) would be to work from the opposite direction - how many themes in the myths of a particular group don't seem to be found elsewhere? That would tend to argue against some kind convergent evolution, at least, but might not rule out inconsistent transmission to other groups. Even in the article, it looks like the data could be used to emphasize the outliers and not the common thread (which does seem to potentially suffer from selection bias).