According to our DRD3 hypothesis, gly is dominant with respect to ser in regard to blocking of the A trait. So the gly/gly and gly/ser genotypes would correspond to N or NP types.
But (again, according to our hypothesis) ser/ser is a necessary but not sufficient condition for A trait, with the necessary haplotype being more complicated. So, the ser/ser group would contain all of the children with A or A− trait, but could contain some N or NP types as well.
To make any sense of the above study, one would need to be familiar with the NPA model and recognize that NP and NPA− types, although very similar from the point of view of aggressive behavior, are on the opposite sides of the coin with regard to the DRD3 ser/gly genotype.