Population biologists analyzing group selection (natural selection at group levels) love doing studies like this. They would generally predict altruistic behavior to protect the females of reproductive age, and would also predict altruistic behavior to favor close relatives, protecting the group's collective genetic makeup. Of course it doesn't work here because of arbitrary groupings, small sample size and retrospective analysis instead of properly designed experiments. But it just struck me that this is what the guy was hoping he could do, I think. "Culture" is an advanced overlay, compared to animal populations which are often social/societal.