Trond Engen sent me links to “Ancient Roots of Indo-European Folktales,” by Sara Graça da Silva and Jamshid J. Tehrani (“In this study, we introduce new methods for tackling these problems by applying comparative phylogenetic methods and autologistic modelling to analyse the relationships between folktales, population histories and geographical distances in Indo-European-speaking societies”), and the (inevitably breathless) BBC News story about it, “Fairy tale origins thousands of years old, researchers say” (“Analysis showed Beauty And The Beast and Rumpelstiltskin to be about 4,000 years old”). My immediate reaction is skepticism, and Trond says:
Maybe I don’t understand their probabilities, but it seems like they use the statistical procedure to seek out the outliers and then say “I’m almost certain that these are cherries!” I mean, 1/3 of almost 300 adventures showing more vertical affinity than expected by chance. Really? Gradually fewer of those 100 adventures being reconstructible (with a low bar) for each node up, resulting in only one out of 100 being (weakly) assigned to the top node. What do they expect from a scattershot? And no attempt to control with e.g. Finnish or Hungarian.
But both of us are eager to hear what Haters have to say about it.