Power for the bootstrap (black) and direct (gray) tests for data based on the standard CD4 dataset (solid), dataset with double the number of subjects (short dash), and dataset with double the observations per subject (long dash). The vertical dashed line indicates the effective power of the tests, where data is simulated directly from the estimated alternative covariance (⁠⁠). From left to right, the settings for are ⁠, and ⁠.