Twin imitation for antisocial behavior: implications for genetic and family environment research.

When twin pairs influence each other's behavior, observed variance is greater for MZ twins than for DZ twins under at least 1 of 2 conditions: (a) the trait has some heritability and (b) MZ twins influence each other more than do DZ twins. Applied to a trait that has an underlying continuous distribution but is measured as a dichotomy, the presence of reciprocal twin influence predicts that if the base rate for the trait is not exactly 50%, then the prevalence of the trait should differ in MZ and DZ twin pairs. This prediction held for registered criminality in a large twin cohort. Methods of analysis that permit reciprocal twin interaction not only provide better statistical fits to the data but also yield estimates of heritability that agree with adoption data. The results suggest that the genetic influence on registered criminality may be more modest than previously thought.