Increase in power through multivariate analyses

Power to detect genetic and environmental influences increases not only wit h sample size but also with the number of measurements through longitudinal and/or multivariate designs, if those measurements correlate with each oth er. Power simulations are presented for uni- through quadrivariate cases, w ith differing genetic and environmental parameters. Even though subject att rition is a problem for most longitudinal studies, the gain in power availa ble may more than make up for this shortcoming in many situations. In terms of planning studies to examine genetic and environmental influences, power calculations should not only consider sample size but number of measuremen ts on particular phenotypes and their intercorrelations.