rmcclosk
commented
8 years ago Did this with alpha1 = 0.5 and alpha2 = 1.5. Perhaps predictably, the result had high posterior density around alpha = 1. This is an important caveat: the model gives us an average of peoples' behaviour with respect to preferential attachment, but this isn't something we can say about any individual person.
Try inference on one network where half the nodes were added with alpha = alpha1, and the other with alpha = alpha2.