e know the real distribution of the F statistic in linear models — it is a non-central F distribution. Under H0, we have a central F distribution. Given 1 – α, we can compute the probability of (correctly) rejecting H0. I created a simple demo to illustrate how the power changes as other parameters vary,