To ensure specificity of effects, a number of demographic, familial and individual-level risks were included as control variables. In addition to offspring sex and age at interview, twin sex, and twin age and education as of the 1997–2002 interview were included among demographic control variables. Dummy variables for having not finished high school and completion of any tertiary education were computed, with high school only comprising the reference group. Control variables for parent comorbid psychopathology include twin and biological coparent history of DSM-IV major depressive disorder (MDD) and a nondiagnostic measure of antisocial personality disorder (ASP). MDD and ASP were assessed of twins in the 1997–2002 interview and both twins and coparents in parent interview. For coparent MDD and ASP, two dummy variables were coded to distinguish affected coparents from those with missing data (11% each for MDD and ASP). Twin history of regular smoking was also assessed in the 1997–2002 interview, with coparent regular smoking queried in parent and offspring interviews. Control variables for offspring comorbid psychopathology include DSM-IV conduct disorder, MDD, social phobia, generalized anxiety disorder, and suicidality
