not shown). Note that these results apply to association testing and not to individual prediction, which is discussed in the next subsection. For association testing there is a balance to be made between the precision of estimating the score in the training subset, and the power of testing the score in the replication subset. For prediction, however, the size of the replication subset does not affect the accuracy, only how precisely it is estimated; thus a larger training subset is more desirable in the prediction context.
