Figure 5.

Sample size as a function of ${\mathit{\alpha }}_1.$ Sample size is computed for $0.5\le {\alpha }_1\le 3$ with 80% power and type-I error rate 0.00025. (A) The sample size when there is a small correlation between the ancestry and other covariates ($r_u^2=0.2$). (B) The sample size when the ancestry is independent of the covariates ($r_u^2=0$).