The results we have shown here are for a given sample size and range of effect sizes. Since power depends on both of these factors in a simple way, they are also more generally applicable. Specifically, the noncentrality parameter is proportional to Nθ2, where θ is the parameter of interest (see Equations (A.3)–(A.5) in Appendix A). For example, if one is interested in what happens for a sample size of 2N, then the same qualitative results would be obtained for as were obtained for θ with sample size N. Thus, it is sufficient to conduct simulations for only one sample size to yield conclusions that hold more generally.
