It seems that the criterion for the selection is how theoretically informative the clusters are. I can imagine that many choices are justifiable, so how to choose and justify the clustering in practice?
I also notice a contradiction between the orienting reading and this paper that I do not know how to handle. If I were doing the clustering, I would probably miss out some important and interesting "surprises" that the algorithms detect because I choose the clusterings that fit with the theoretical framework that I bring into the study.
It seems that the criterion for the selection is how theoretically informative the clusters are. I can imagine that many choices are justifiable, so how to choose and justify the clustering in practice? I also notice a contradiction between the orienting reading and this paper that I do not know how to handle. If I were doing the clustering, I would probably miss out some important and interesting "surprises" that the algorithms detect because I choose the clusterings that fit with the theoretical framework that I bring into the study.