Thank you for the citations and leads to some very creative people!
You mentioned "variance." A large part of my concern in "Here Comes Grand Dad Again" is that group data sometimes hide the most interesting phenomena. (Skinner once raised a similar point.) For example, learning appears to be a binary experience but a group of critters will show a curve as different numbers of them catch on at different times. RA Fisher (and others) did the social sciences a large favor by introducing analytical techniques for group data. His contribution also nudged research down one evolutionary path for assessing facts; yet, Sherrington, Pavlov, Katz, Sperry, and even Freud followed another and made their contributions through language about individuals, not about means and SDs.
Most (all?) of the genetic studies in mental health look for clusters of diagnoses, rather than detailed mannerisms. I am unsure about the extent of work that examines mannerisms for several generations in order to shed indirect light on behavior patterns that alternate generations because of the possible influence of allele crossover during meiosis.
My own experiences are self-selected bits of data with lots of interviewer bias; thus, none qualifies as lab science. Most clinicians I know have observed similar phenomena, in their own or in their clients' lives. No one has good explanations, especially for behaviors that skip generations; indeed, no one in my circle has considered them to be significant events. On the other hand, parents usually relax when their child's aberrant conduct is noted to be similar to their own or to their parents' (both in earlier years and in more refined expression in adulthood). There's a glimmer of "it makes sense" and some relaxing. Mom & dad usually feel that they didn't do so badly; their nightmares about their child usually diminish. Most important, their understanding appears to grow about the need for parental consistency as well as to set goals and to reduce the impact of any social or academic "disabilities" by finding areas of talent in their child.
NOTE:
It's also intriguing as to why more people don't make systematic study of these behavioral snippets. Some Complex Adaptive Systems let Gould (but certainly not me!) classify snails; it needs to be done for vocational talents or the detailed movements in a walk or in a style of courting. One difference may relate to the data. Although both realms are observable, we can do more "parallel processing" comparisons (looking at lots of samples at once) with snails than we can with shoplifting or phobias. Perhaps our data and our explanations about the sources of our behavioral features are too limited by the operating traits of our Adaptive Systems for noticing and classifying facts. (Fisher's methods were for evaluating the probability of a difference between groups; we have Adaptations for noticing "identical" or "similar" events in daily life and we will need to use them in our behavior analyses.)