Hm, does not appear sound to me because it seems to be utterly impossible to control for other factors that happened during the last 2k years. I completely agree that there may be the possibility to segment the population along an imaginary line that matches the extend of the Roman Empire. But to conclude that this segmentation was caused by a state border 2k years ago needs extraordinary evidence. Some statistical analysis is not enough to verify or reject this assumption.