COGA sample. Unlike older adult samples previously studied in COGA, there is no association at the level of diagnostic clinical symptoms in this younger sample. Instead, the association is manifest at this stage of development with subclinical levels of externalizing behavior that do not reach the level of diagnostic symptoms until later adulthood. This is in contrast to a previous COGA publication analyzing a smaller sample (N = 850) of children/adolescents, in whom we found evidence of association with conduct disorder symptoms (Dick et al., 2006b). This sample differs in two important ways that may have contributed to the discrepant findings with respect to an association between GABRA2 and clinical level conduct disorder symptoms. In the sample in which an association was found, 77% of children were part of the most densely affected families in COGA; this prospective sample follows all children from both COGA proband and comparison families. That sample also combined information about conduct disorder across multiple assessments and used the maximal symptom count score from any assessment; it is possible that this led to a more reliable and/or more severe conduct disorder symptom count.
