estimates for a significance level α and a power value β using non-parametric regression.

<p>These values were averaged over ten simulations each. Note that these values differ slightly from those obtained by Samuels et al. (see their Table 3) <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0073567#pone.0073567-Samuels1" target="_blank">[26]</a> due to the simulation procedure implemented in both studies and because we consider unequal number of cases and controls.