Modeling the association of school achievement and criminal behavior in Swedish twins, full & half siblings, and cousins using SEM and co-relative designs

Using Swedish nationwide registry data, we compared testing the contribution of genetic and environmental risk factors to the association of school achievement and criminal behavior by structural equation modeling and by... [ view full abstract ]

Using Swedish nationwide registry data, we compared testing the contribution of genetic and environmental risk factors to the association of school achievement and criminal behavior by structural equation modeling and by co-relative control designs.

School achievement data were obtained from conscription registries while criminal behavior was derived from criminal records. Twin, full sibling, half sibling, and cousin pairs born between 1960-1990 were obtained from the national twin and genealogical registers. Genetic structural equation modeling and logistic regression were applied to the population-based data, using OpenMx and R.The focus of SEM modeling was more empirically driven by fitting all possible identified bivariate models including up to three parameters accounting covariance from additive genetic, shared environmental and unique environmental sources and reciprocal causal paths, with simultaneous testing for qualitative and quantitative sex differences.  Analyses were done separately for twin pairs only, including twins & full siblings, adding half sibling, testing just full & half siblings and finally combining twins, full & half sibs and cousins to evaluate the power of alternative designs.  Furthermore, we tested the information gained from SEM modeling compared to using discordant co-relative designs.