analyses should be in light of the twin and sibling sample used, as there is some residual uncertainty about whether inferences from twin data have external validity with respect to what might be found in general population samples (Vitaro et al., 2009). Analyses were adjusted for clustering effects using the Huber-White estimator, which was selected over other potential analyses that can be conducted to explore within twin/sibling frailties as the most parsimonious method.
