P(D) is the probability of observing the data you just saw, due to either the null or non-null hypothesis. It's a strictly Bayesian quantity, since it's depende

P(D) is the probability of observing the data you just saw, due to either the null or non-null hypothesis. It's a strictly Bayesian quantity, since it's dependent on a prior. If your model has only a null and alternative hypothesis, then: P(D) = P(D|H0)P(H0) + P(D|H1)P(H1)