Sex was entered as a covariate in view of previous sex differences for this outcome (Newman et al., 1996). We also included the number of reporters on the parental antisocial behavior and parental alcohol problems measures (i.e., 1 parent or 2 parents) as covariates in the respective models in order to account for the possibility that missing parental data, particularly missing father data, may reflect some degree of risk for externalizing behavior.
