A Note on the Statistical Power in Extended Twin Designs
- 268 Downloads
The power to detect sources of genetic and environmental variance varies with sample size, study design, effect size and the statistical significance level chosen. We explored whether the power of the classical twin study may be increased by adding non-twin siblings to the classical twin design. Sample sizes to detect genetic and shared environmental variation were compared for kinships with only twins, kinships consisting of twins and one additional sibling, and kinships with twins and two additional siblings. The effect of adding siblings to the classical twin design was considered for univariate and bivariate analyses. For the univariate case, adding one non-twin sibling resulted in a decrease in sample size needed to detect additive genetic influences in the presence of environmental influences. However, adding two additional siblings did not decrease the number of subjects as compared to the classical twin design. The sample size required to detect common environmental factors was also greatly decreased by adding one non-twin sibling. Adding two non-twin siblings resulted in a small additional decrease. In models including additive genetic, dominant genetic, and unique environmental effects, adding one sibling to a twin family decreased the required sample size to detect dominant genetic influences. Adding two siblings to a twin family resulted in only a slight additional decrease in sample size. In the bivariate case a similar pattern of results was found, in addition to the observation that the overall required sample size, as expected, was lower than in the univariate case. The decrease in sample size from bivariate testing was more pronounced in a design with one or two additional siblings, as compared to a design with twins only. It is concluded that a well considered choice of family design, i.e. including families with twins and one or two additional siblings increases the statistical power to detect sources of variance due to additive and non-additive genetic influences, and common environment.
Unable to display preview. Download preview PDF.
- Boomsma, D. I. Current status and future prospects in twin studies of the development of cognitive abilities: Infancy to old age. In: Bouchard, Thomas J. Jr. (Ed), Propping, Peter (Ed), et al. (1993). Twins as a tool of behavioral genetics.Life sciences research report, 53. (pp. 67–82). Chichester, England UK: John Wiley & Sons.Google Scholar
- Cohen, J. (1992) A power primer. Psychological Bulletin, 112:155–159.Google Scholar
- Dolan, C. V., Boomsma, D. I., & Neale, M. C. (1999). A note on the power provided by sibships of size 2, 3, and 4 in genetic covariance modeling of a codominant QTL. In press Google Scholar
- Neale, M. C. (1997). Mx: Statistical modeling. 3rd edition Box 980126 MCV, Richmond VA 23298.Google Scholar
- Neale, M. C., & Cardon L. R.(1992) Methodology for Genetic Studies of Twins and Families. NATO Asi Series. Series D,Behavioural and Social Sciences, Vol 67.Google Scholar
- Plomin, R., DeFries, J. C., & McClearn, G. E. (1990) Behavioral Genetics. A primer. New York: Freeman and company.Google Scholar
- Svikis, D. S., Velez, M. L., Pickens, R. W. (1994). Genetic aspects of alcohol use and alcoholism in women. Alcohol Health & Research World, 18:192–196.Google Scholar
- Tanaka, J. S. (1987). How big is big enough?: sample size and goodness of fit in structural equation models with laten variables. Child Development, 58:134–146.Google Scholar