Peer deviance (PD) is associated with risk for drug abuse (DA). Is this association causal?

DA was recorded in official records. PD was defined as the percentage of peers residing in small communities with future DA registrations. We examined offspring in families whose community PD changed when the offspring was 0–15 years of age and then examined families where cousins or siblings differed in their years of exposure to low or high PD communities.

The duration of exposure to PD was strongly associated with future DA. Co-relative analyses for families whose exposure to PD declined suggested that the PD–DA association was largely non-causal. Within full-sibling pairs in such families, the length of exposure to low PD environments was unrelated to risk for DA. By contrast, co-relative analyses in families where exposure to PD increased over time indicated that the PD–DA association was largely causal. In such families, siblings who differed in the duration of their exposure to high PD differed in their risk for subsequent DA. These results were replicated in families whose PD changed because they moved or because of changes in the community in which they resided.

Within families whose social environment is improving over time, the association between PD exposure and offspring DA outcomes is not causal but is due to familial confounding. Within families whose social environment is deteriorating, the PD–DA association seems to be largely causal. Our measure of PD may also reflect broader aspects of the community environment beyond peers.

Conduct disorder (CD) and peer deviance (PD) both powerfully predict future externalizing behaviors. Although levels of CD and PD are strongly correlated, the causal relationship between them has remained controversial and has not been examined by a genetically informative study.

Levels of CD and PD were assessed in 746 adult male–male twin pairs at personal interview for ages 8–11, 12–14 and 15–17 years using a life history calendar. Model fitting was performed using the Mx program.

The best-fit model indicated an active developmental relationship between CD and PD including forward transmission of both traits over time and strong causal relationships between CD and PD within time periods. The best-fit model indicated that the causal relationship for genetic risk factors was from CD to PD and was constant over time. For common environmental factors, the causal pathways ran from PD to CD and were stronger in earlier than later age periods.

A genetically informative model revealed causal pathways difficult to elucidate by other methods. Genes influence risk for CD, which, through social selection, impacts on the deviance of peers. Shared environment, through family and community processes, encourages or discourages adolescent deviant behavior, which, via social influence, alters risk for CD. Social influence is more important than social selection in childhood, but by late adolescence social selection becomes predominant. These findings have implications for prevention efforts for CD and associated externalizing disorders.