a very small effect individually and might therefore not be detected in the analyses [30], [47]. We estimated that the total proportion of phenotypic variance explained by genome-wide SNPs when considered together is 0.55, with a standard error of 0.41. The point estimate is non-significantly different from zero and larger sample sizes, enriched for cases, will be required to ensure sufficient power to accurately estimate the proportion of phenotypic variance in adult antisocial behavior explained by all the genome-wide SNPs. The application of this methodology to criminal behavior phenotypes is particularly relevant, a field in which a genetic contribution to the etiology remains contentious. Although the classical twin design for estimation of heritability is designed to separate out the common family environment effects from genetic effects in the familial relationship, some contamination with common environmental effects could remain [48]. The methods of Yang et al, estimate the contribution of genetic effects from such distantly relatives that contamination with family environmental effects is less likely.
