In my opinion, the weakness of this line of reasoning is that there are a _lot_ of socially maladjusted kids and young adults, and very few of them become mass murderers.
So, the observation that a large fraction of actual killers fit this type has little or no predictive value.