Modeling extended twin family data II: power associated with different family structures

Modeling the data from extended twin pedigrees allows the estimation of increasing complex covariance relationships in which the effects of cultural transmission, nonrandom mating and genotype x environment covariation can be incorporated. However, the power to detect these effects in existing data sets has not yet been examined. The present study examined the effects that different family structures (i.e., the ratio of MZ to DZ families and the importance of cousins vs. avuncular relatives) have on statistical power. In addition, we examined the power to detect genetic and environmental effects within the context of two large data sets (VA30K and the OZVA60K). We found that power to detect additive genetic and cultural transmission effects were maximized by over sampling MZ families. In terms of ascertainment, there was little difference in power between samples that had focused on recruiting a third generation (the children of twins) versus those that had focused on recruiting the siblings of the twins. In addition, we examined the power to detect additive and dominant genetic effects, cultural transmission and assortative mating in the existing VA30K and OZVA60K samples, under two different models of mating: phenotypic assortment and social homogamy. There was nearly 100% power to detect assortative mating and cultural transmission, against a background of small additive and dominant genetic and familial environmental effects. In addition, the power to detect additive or dominant genetic effects quickly asymptoted, so that there was almost 100% power to detect effects explaining 20% or more of the total variance. These results demonstrate that the Cascade model has sufficient power to detect parameters of interest in existing datasets. Mx scripts are available from www.vipbg.vcu.edu/~sarahme/cascade.